Financial Risk Specialists

13-2054.00
Median wage $117,330/yr63,850 employed (US)Rank #80 of 923 scored · top 9% by substitution

Analyze and measure exposure to credit and market risk threatening the assets, earning capacity, or economic state of an organization. May make recommendations to limit risk.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution50
Exposure48
Augmentation85

Substitution — the headline: capability discounted by cost, barriers and adoption.

Exposure — technical capability alone, regardless of whether anyone deploys it.

Augmentation — how much AI assists without replacing. High here + moderate substitution = a changing job, not a disappearing one.

Tasks on the substitution scale

30 rated tasks, binned by substitution score.

Position among all scored occupations

Distribution of 923 occupation scores; the marker is this occupation.

Tasks with substitution ≥ 70

10%

Run 1.0.0-draft.1 · computed 2026-08-05 · rater panel: claude-sonnet-5, claude-haiku-4-5-20251001 · intervals span rater disagreement.

Why this score

The five weighted dimensions of the composite, averaged across this occupation's tasks (importance-weighted, panel mean). Exact weights and formulas: /api/v1/methodology.

Task automatabilityw 35%50

panel mean rating 3.0/5 → substitution pressure 50/100

Technical feasibility todayw 20%45

panel mean rating 2.8/5 → substitution pressure 45/100

Cost vs. human wagew 15%54

panel mean rating 3.2/5 → substitution pressure 54/100

Adoption barriersw 20%inverted — strong barriers lower the score46

panel mean rating 3.1/5 (barrier strength) → substitution pressure 46/100

Sector adoption velocityw 10%61

panel mean rating 3.4/5 → substitution pressure 61/100

Task breakdown (30 tasks)

Substitution pressure per task, weighted by O*NET importance in the composite. Expand a task for the full rater audit trail — every rating, every model, every rationale.

Draw charts and graphs, using computer spreadsheets, to illustrate technical reports.

95

CI 92–97 · exposure 100 · augmentation 100 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Financial services and professional services sectors are rapidly adopting AI-assisted analytics and visualization tools. Spreadsheet-based charting automation is common in fintech, banking, and corporate strategy teams, with production deployment at scale across major firms.
Sector adoption velocityclaude-sonnet-54/5Financial services is a fast-adopting sector for AI tools, and spreadsheet/BI charting automation is already widely used in production workflows.
Augmentation potentialclaude-haiku-4-5-202510015/5AI dramatically augments human productivity in this task: specialists can iterate chart designs, regenerate for different data slices, and refine aesthetics instantly. The human retains judgment over visualization strategy and message, but AI handles the mechanical work at transformative speed gains.
Augmentation potentialclaude-sonnet-55/5AI significantly speeds up and improves chart creation for technical reports, letting analysts focus on interpretation rather than manual formatting.
Task automatabilityclaude-haiku-4-5-202510015/5Current AI systems (e.g., Claude, GPT-4, specialized charting tools) can fully automate chart and graph generation from spreadsheet data with substantial time savings. Users can describe requirements and AI produces publication-ready visualizations with minimal human intervention, easily exceeding 50% time savings at equal quality.
Task automatabilityclaude-sonnet-55/5Generating charts and graphs from data in spreadsheets is a well-defined, formulaic task that current AI tools (e.g., Excel Copilot, Python code generation, chart-generation LLMs) can do end-to-end with substantial time savings at equal or better quality.
Adoption barriersclaude-haiku-4-5-202510012/5Few hard barriers exist: charting is not a licensed or regulated task, no legal sign-off is required, and organizations face minimal liability for chart errors in technical reports. Light oversight (human review of output) is typical but not a legal requirement, creating only modest friction.
Adoption barriersclaude-sonnet-51/5No licensing, liability, or regulatory barrier prevents using software or AI to create charts; this has been standard practice for decades.
Cost vs. human wageclaude-haiku-4-5-202510015/5AI inference cost for generating charts is negligible (often <$0.01 per chart), while a financial risk specialist's loaded hourly wage (typically $75–150+/hour) makes human-generated charting far more expensive. AI is orders of magnitude cheaper.
Cost vs. human wageclaude-sonnet-55/5Automated chart generation costs fractions of a cent in compute versus analyst time, making AI at least an order of magnitude cheaper for this narrow task.
Technical feasibility todayclaude-haiku-4-5-202510015/5Mature deployed products for automated chart generation exist at scale: Excel/Google Sheets native AI features, Tableau, Power BI, and LLM-based tools reliably produce charts from data in production environments. Error rates on standard chart types are low and the technology is widely adopted.
Technical feasibility todayclaude-sonnet-55/5Mature deployed products (Excel/Google Sheets AI features, BI tools like Tableau with NL-to-chart, Copilot for Office) reliably generate charts from data in production today.

Track, measure, or report on aspects of market risk for traded issues.

74

CI 53–95 · exposure 75 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510015/5Financial services is among the fastest-adopting sectors for AI automation; market risk measurement and algorithmic reporting are already deeply embedded in trading and risk management workflows across investment banks and hedge funds.
Sector adoption velocityclaude-sonnet-54/5Financial services is among the fastest-adopting sectors for AI/quant tools, with risk analytics and reporting automation already widespread in trading and risk management functions.
Augmentation potentialclaude-haiku-4-5-202510014/5AI systems augment risk specialists by automating metric computation, enabling faster scenario analysis, stress testing, and real-time monitoring; humans remain in the loop for interpretation, governance decisions, and exceptional risk events.
Augmentation potentialclaude-sonnet-54/5AI substantially speeds up data gathering, metric computation, anomaly flagging, and draft report generation, letting risk specialists focus on judgment and escalation while staying in the loop.
Task automatabilityclaude-haiku-4-5-202510015/5Tracking and measuring market risk for traded instruments involves quantitative calculation, historical data analysis, and standardized reporting—core strengths of current AI systems. Modern tools can compute Value-at-Risk, volatility metrics, correlation analysis, and generate risk reports end-to-end with substantial time savings and equal quality.
Task automatabilityclaude-sonnet-53/5AI can pull data, compute standard risk metrics (VaR, duration, beta) and draft reports, but interpreting model limitations, tail scenarios, and reporting judgment calls still require human oversight, so only partial time savings are realized end-to-end.
Adoption barriersclaude-haiku-4-5-202510012/5While regulatory oversight and sign-off requirements exist in financial services, the computation and measurement itself can be fully automated; human review and sign-off occur downstream. No licensing requirement explicitly mandates human performance of the technical calculation itself.
Adoption barriersclaude-sonnet-53/5Market risk reporting for traded issues is subject to regulatory scrutiny (e.g., Basel, SEC) and often requires sign-off by a qualified risk officer, creating moderate friction even though the underlying calculations can be automated.
Cost vs. human wageclaude-haiku-4-5-202510015/5The per-report cost of automated risk calculation and measurement is several orders of magnitude cheaper than paying a human analyst or risk specialist to manually compute the same metrics across multiple instruments and time horizons.
Cost vs. human wageclaude-sonnet-53/5Automated risk calculation engines reduce headcount needs somewhat, but licensing, data feeds, model validation, and compliance oversight keep total cost roughly comparable to skilled analyst labor rather than an order of magnitude cheaper.
Technical feasibility todayclaude-haiku-4-5-202510015/5Multiple production systems in financial institutions already perform market risk tracking and measurement via algorithmic analysis, backtesting engines, and automated dashboards. Banks and trading firms deploy these systems at scale for daily risk reporting and compliance.
Technical feasibility todayclaude-sonnet-53/5Risk analytics platforms and AI-assisted reporting tools are deployed in banks and asset managers, but they typically automate data aggregation and metric calculation rather than the full measurement-and-reporting workflow, and require human validation before reports are finalized.

Gather risk-related data from internal or external resources.

72

CI 70–75 · exposure 75 · augmentation 100 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Financial services firms are rapidly deploying RPA and AI for data aggregation and risk reporting; adoption is measurably deep in large institutions and accelerating in mid-market operations.
Sector adoption velocityclaude-sonnet-54/5Financial services is a fast-adopting sector for data automation and AI tooling, with widespread use of automated data feeds and analytics pipelines already in production.
Augmentation potentialclaude-haiku-4-5-202510015/5AI assistants substantially boost specialist productivity by pre-aggregating data, flagging anomalies, and formatting feeds for analysis, allowing humans to focus on interpretation and decision-making rather than manual collection.
Augmentation potentialclaude-sonnet-55/5AI tools substantially speed up and broaden data collection, letting risk specialists focus on analysis and interpretation rather than manual retrieval.
Task automatabilityclaude-haiku-4-5-202510014/5AI systems can reliably extract, aggregate, and retrieve risk data from structured databases, APIs, and documents with high efficiency, easily exceeding 50% time savings. However, validation of data quality and judgment about source reliability may still require human oversight in complex or novel scenarios.
Task automatabilityclaude-sonnet-54/5Data gathering from structured databases, APIs, and documents is highly automatable using scripts, RPA, and AI-driven extraction tools, though some sources require judgment about relevance and quality.
Adoption barriersclaude-haiku-4-5-202510013/5Data governance, compliance (e.g., data residency, audit trails), and internal authorization protocols create moderate friction. Financial regulations require documented data provenance and human sign-off on critical risk datasets, slowing pure automation.
Adoption barriersclaude-sonnet-52/5No licensing requirement to gather data, though internal data governance, access permissions, and confidentiality agreements create moderate friction.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI-driven data gathering (automated web scraping, API calls, document parsing) costs a fraction of a specialist's time per data point collected; the cost advantage is substantial when amortized across recurring data pulls.
Cost vs. human wageclaude-sonnet-54/5Automated data pulls and extraction are dramatically cheaper than manual gathering, though integration and validation add some ongoing cost.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed products (data aggregation platforms, RPA tools, LLM-based document parsing) perform risk data gathering at scale in financial institutions. Some limitations remain in handling unstructured or proprietary formats, but core functionality is production-ready.
Technical feasibility todayclaude-sonnet-54/5Deployed data aggregation, web scraping, and financial data platforms (e.g., Bloomberg APIs, RPA tools, AI document extraction) reliably gather risk data in production today, though full breadth of unstructured or proprietary sources still needs some human curation.

Consult financial literature to ensure use of the latest models or statistical techniques.

70

CI 61–79 · exposure 62 · augmentation 100 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Financial services, asset management, and risk quantitative teams are rapidly adopting AI-assisted research tools, document automation, and literature synthesis. Major banks and hedge funds actively deploy AI for research aggregation and have moved past pilot stage.
Sector adoption velocityclaude-sonnet-54/5Financial services is a fast-adopting sector for AI-based research and knowledge tools, with widespread pilot and production use of AI search/summarization in risk and quant teams.
Augmentation potentialclaude-haiku-4-5-202510015/5AI dramatically augments a financial risk specialist's productivity by automating the discovery, filtering, and summarization of vast literature while the human retains judgment on applicability and model selection. This is a canonical augmentation scenario where AI handles information overload and the specialist provides expert curation.
Augmentation potentialclaude-sonnet-55/5AI substantially speeds up literature discovery, summarization, and cross-referencing of statistical techniques, letting specialists cover far more ground while retaining judgment over final model selection.
Task automatabilityclaude-haiku-4-5-202510014/5Literature review—identifying, retrieving, and synthesizing recent financial models and techniques—is largely automatable by AI systems that can search academic databases, parse papers, and summarize findings. However, the critical judgment of which models are genuinely 'latest' or applicable to a specific firm's risk context typically requires human expertise, preventing full end-to-end replacement while still saving >50% of the reading and initial synthesis work.
Task automatabilityclaude-sonnet-53/5AI can search, summarize, and synthesize financial literature and identify relevant models/techniques quickly, but validating applicability and correctness for firm-specific risk contexts still requires human judgment, so only partial time savings accrue end-to-end.
Adoption barriersclaude-haiku-4-5-202510012/5No regulatory, licensing, or legal requirement mandates human literature review; IP and confidentiality concerns are minimal for reviewing published research. The main friction is organizational conservatism and analyst preference to maintain control over research input, not formal barriers.
Adoption barriersclaude-sonnet-52/5No licensing requirement mandates a human perform literature review itself, though downstream model validation and risk sign-off may carry regulatory expectations that indirectly create some organizational caution.
Cost vs. human wageclaude-haiku-4-5-202510015/5AI-assisted literature review (automated searches, summaries, and relevance filtering) costs a small fraction of the analyst time it would take to manually read and synthesize the same volume. A few dollars in API costs replaces hours of specialist labor at $100+/hour loaded cost.
Cost vs. human wageclaude-sonnet-54/5AI-assisted literature search and summarization is dramatically cheaper than having a specialist manually review journals and papers, though some human review cost remains to verify accuracy.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed AI systems (LLMs, semantic search, arXiv/journal aggregation tools, and document analyzers) reliably perform literature triage, summarization, and technique extraction at scale in production financial and research environments. Error rates on factual extraction are low, though human verification of technical claims remains standard practice.
Technical feasibility todayclaude-sonnet-53/5Deployed tools (e.g., AI research assistants, literature-summarization products, financial data platforms with LLM search) exist and are used in practice, but they have material error rates on technical nuance and are not fully trusted for authoritative literature review without human vetting.

Monitor developments in the fields of industrial technology, business, finance, and economic theory.

66

CI 61–71 · exposure 55 · augmentation 100 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Financial services, asset management, and fintech firms are actively deploying AI-driven market intelligence, news aggregation, and risk-alert systems. Adoption is visible and accelerating in large institutions; smaller firms are in pilot phases, consistent with mature-sector adoption.
Sector adoption velocityclaude-sonnet-54/5Financial services is a fast-adopting sector for AI research and monitoring tools, with many firms already using AI-powered market and news intelligence platforms.
Augmentation potentialclaude-haiku-4-5-202510015/5AI significantly augments financial risk specialists by enabling real-time monitoring across vastly larger information spaces, faster anomaly detection, and dynamic alert prioritization—all while the specialist retains judgment on materiality, strategy, and enterprise context.
Augmentation potentialclaude-sonnet-55/5AI tools excel at continuously scanning, summarizing, and alerting on developments across fields, significantly boosting the productivity of a human monitoring these domains.
Task automatabilityclaude-haiku-4-5-202510013/5AI can automatically ingest, filter, and summarize news and research across industrial, business, finance, and economic domains at scale, meeting the 50% time-saving bar for the information-gathering and synthesis portion. However, contextualizing developments within an organization's specific risk profile and determining materiality still require human judgment, limiting full end-to-end automation.
Task automatabilityclaude-sonnet-53/5AI can aggregate, summarize, and flag relevant news/research on industry, finance, and economic developments, but synthesizing implications for firm-specific risk still requires human judgment, limiting full end-to-end automation.
Adoption barriersclaude-haiku-4-5-202510012/5Monitoring developments is predominantly an information task with no hard legal requirement for human execution. Financial firms are adopting AI monitoring tools readily; oversight and human sign-off on *material* risks remain standard, but the monitoring itself faces minimal regulatory or liability barriers.
Adoption barriersclaude-sonnet-51/5This is an informational/monitoring task with no licensing or sign-off requirement; there is little to prevent AI-assisted or AI-driven monitoring.
Cost vs. human wageclaude-haiku-4-5-202510014/5Automated monitoring via subscription-based AI platforms (search, aggregation, summarization) costs a fraction of a specialist's fully loaded salary for equivalent information coverage. One system can monitor for dozens of specialists; the cost leverage is significant.
Cost vs. human wageclaude-sonnet-54/5AI-driven monitoring and summarization tools are dramatically cheaper than dedicating analyst hours to continuously scan multiple fields, though some oversight cost remains.
Technical feasibility todayclaude-haiku-4-5-202510014/5Production systems (financial news aggregators, AI-powered research platforms, trading-desk alert tools) reliably monitor and summarize developments across these fields in real time. Some error in relevance filtering and interpretation exists, but deployed products demonstrably perform core monitoring functions at scale in financial institutions.
Technical feasibility todayclaude-sonnet-53/5News aggregators, AI summarization tools, and research assistants are deployed in finance today, but they still miss nuance, have hallucination risk, and require human curation to be reliable.

Evaluate and compare the relative quality of various securities in a given industry.

64

CI 53–76 · exposure 62 · augmentation 100 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510015/5Quantitative finance, asset management, and fintech firms have rapidly deployed algorithmic security ranking and comparison tools. Major asset managers, hedge funds, and institutional investors are already using ML-driven comparative scoring in production as a core part of investment workflows.
Sector adoption velocityclaude-sonnet-54/5Financial services is among the fastest-adopting sectors for AI-driven research and analytics, with widespread pilot-to-production deployment of AI research assistants and comparative analytics tools.
Augmentation potentialclaude-haiku-4-5-202510015/5AI dramatically augments human analysts by instantly surfacing data, computing metrics across hundreds of securities, and highlighting outliers, freeing specialists to focus on forward-looking context, industry dynamics, and judgment—a classic high-augmentation, partial-automation scenario.
Augmentation potentialclaude-sonnet-55/5AI substantially augments this task by rapidly aggregating financial statements, market data, and peer comparisons, letting analysts focus on judgment and synthesis rather than manual data gathering.
Task automatabilityclaude-haiku-4-5-202510014/5AI can analyze financial statements, calculate key metrics, and produce comparative reports on securities across quantitative dimensions (valuation ratios, growth rates, credit metrics) at high speed, achieving >50% time savings. However, subjective quality judgments, forward-looking analysis, and industry-specific context still require human oversight and interpretation.
Task automatabilityclaude-sonnet-53/5AI can gather and synthesize financial data, ratios, and qualitative factors to compare securities, but final judgment involving market context, risk appetite, and non-quantifiable factors still requires human oversight, limiting full automation.9,999 words placeholder removed.9,999 words placeholder removed.9,999 words placeholder removed.9,999 words placeholder removed.9,999 words placeholder removed.
Adoption barriersclaude-haiku-4-5-202510013/5Regulatory oversight exists (SEC oversight of investment advice, compliance requirements) but does not legally require human sign-off for internal comparative analysis. However, institutional risk tolerance, liability concerns for flawed rankings, and organizational culture around human judgment create meaningful adoption friction.
Adoption barriersclaude-sonnet-53/5While no license is strictly required to run comparative analysis, regulatory expectations (e.g., suitability, fiduciary duty) and firm compliance frameworks create moderate friction requiring human sign-off on investment recommendations.
Cost vs. human wageclaude-haiku-4-5-202510015/5Algorithmic comparison of securities costs orders of magnitude less than human analyst time—a single API call and model inference can replace hours of manual data gathering, calculation, and report assembly. Maintenance and licensing are fixed costs spread across thousands of comparisons.
Cost vs. human wageclaude-sonnet-53/5AI tools reduce research time significantly, but licensing costs for data feeds, model outputs, and required human review keep costs roughly comparable to analyst time rather than an order of magnitude cheaper.
Technical feasibility todayclaude-haiku-4-5-202510014/5Multiple financial platforms (Bloomberg Terminal integrations, FactSet, Refinitiv, quant platforms) now offer automated comparative scoring and ranking of securities with real-time data feeds. These are deployed at scale in institutional finance, though they typically support rather than fully replace human judgment on nuanced quality assessments.
Technical feasibility todayclaude-sonnet-53/5Products like Bloomberg AI tools, equity research copilots, and LLM-based analysis platforms exist and are used in production, but comparative quality judgments still show material error rates and require analyst validation.

Conduct statistical analyses to quantify risk, using statistical analysis software or econometric models.

63

CI 53–74 · exposure 62 · augmentation 100 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Financial services and risk management are digitally mature sectors with high adoption of AI-driven analytics platforms. Production deployments of automated statistical modeling and risk quantification are widespread in large banks and asset managers, indicating fast and deep adoption.
Sector adoption velocityclaude-sonnet-54/5Financial services is a fast-adopting sector for AI/ML tools in quantitative analysis, with widespread use of Python-based automation and increasing integration of LLM-assisted coding and analysis in risk functions.
Augmentation potentialclaude-haiku-4-5-202510015/5AI dramatically augments productivity by automating model execution, flagging anomalies, and generating initial interpretations while specialists focus on model design, validation, and strategic risk judgment. This creates substantial augmentation value even where human review remains essential.
Augmentation potentialclaude-sonnet-55/5AI substantially accelerates code generation, model specification exploration, data cleaning, and documentation for statistical risk analyses, letting risk specialists focus on interpretation and judgment while remaining in the loop.
Task automatabilityclaude-haiku-4-5-202510014/5Statistical analysis using econometric models and standard software packages can be largely automated end-to-end with modern AI systems. Setup of model specifications, data preparation, and interpretation remain partly manual, but AI can execute model runs, produce outputs, and flag statistical significance—easily meeting the 50% time-saving threshold for routine quantification work.
Task automatabilityclaude-sonnet-53/5AI can execute standard statistical/econometric analyses (VaR, regression, Monte Carlo) with proper setup, but model selection, data validation, and interpretation of results in context still require significant human judgment, especially for novel or high-stakes risk models.
Adoption barriersclaude-haiku-4-5-202510013/5While statistical analysis itself has few legal barriers, many financial institutions impose governance and sign-off requirements on risk modeling outputs for regulatory or internal control reasons. These organizational oversight requirements and the need for human validation create moderate friction against full automation.
Adoption barriersclaude-sonnet-53/5Financial risk models used for regulatory reporting (e.g., Basel, SR 11-7 model risk management) require documented human validation and sign-off, creating moderate barriers, though the analysis itself isn't inherently restricted to licensed individuals.
Cost vs. human wageclaude-haiku-4-5-202510015/5Computational inference for statistical models is extremely cheap per execution relative to the fully-loaded cost of a financial risk specialist to run analyses manually. Cloud-based modeling and open-source tools make the per-task AI cost orders of magnitude lower.
Cost vs. human wageclaude-sonnet-53/5AI can cut time on coding and running standard models substantially, but licensing for specialized econometric software, compute costs, and mandatory human review of outputs keep all-in costs closer to comparable rather than order-of-magnitude cheaper.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed AI and statistical automation tools reliably execute econometric modeling, regression analysis, and risk quantification in production financial environments. However, interpretation of results and model validation still often require human expertise, preventing a full 5-rating.
Technical feasibility todayclaude-sonnet-53/5Tools like Python/R libraries, AI copilots for code generation, and specialized risk platforms exist and are used in production, but they still require analyst oversight for model validation, assumption-checking, and regulatory compliance rather than fully autonomous operation.

Inform financial decisions by analyzing financial information to forecast business, industry, or economic conditions.

62

CI 53–72 · exposure 62 · augmentation 100 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510015/5Financial services is the fastest-adopting sector for AI; major banks, hedge funds, and asset managers have deployed ML-driven forecasting and risk analysis in production for years. Adoption is deep and accelerating across equities, fixed income, and macro analysis.
Sector adoption velocityclaude-sonnet-54/5Financial services is among the fastest-adopting sectors for AI-driven analytics and forecasting tools, with widespread pilot-to-production deployment of AI copilots in risk and finance functions.
Augmentation potentialclaude-haiku-4-5-202510015/5AI dramatically enhances human financial analysts by processing vast datasets, identifying patterns, and generating scenario analyses in seconds, allowing humans to focus on interpretation, judgment, and strategic recommendation. This is a canonical augmentation case.
Augmentation potentialclaude-sonnet-55/5AI substantially augments this task by rapidly aggregating data, running scenario models, and drafting forecasts, letting human specialists focus on judgment, validation, and strategic interpretation.
Task automatabilityclaude-haiku-4-5-202510014/5Current AI systems can perform large portions of this task end-to-end—gathering financial data, running forecasting models, and generating trend reports—with significant time savings. However, the subjective judgment required for high-stakes business decisions and the need for domain expertise in interpreting nuanced economic signals prevents a full 5 rating.
Task automatabilityclaude-sonnet-53/5AI can rapidly process financial data and generate forecasts, but synthesizing this into decision-informing analysis requires contextual judgment, institutional knowledge, and accountability that current systems only partially replicate.forcing significant human oversight.
Adoption barriersclaude-haiku-4-5-202510013/5Regulatory oversight (SEC, banking regulators) requires human sign-off on material financial advice and risk decisions, creating friction. However, AI augmentation and preliminary analysis are permitted, and liability can be allocated through clear disclaimers and human accountability structures.
Adoption barriersclaude-sonnet-53/5No strict licensing requirement mandates a human perform forecasting, but regulatory scrutiny (e.g., SEC, internal risk controls) and liability for erroneous financial guidance create meaningful friction against full automation.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI-powered financial analysis tools cost a small fraction of specialized human analyst salaries when amortized across many analyses. The per-task inference and integration cost is typically 5–20% of the loaded cost of a qualified risk specialist.
Cost vs. human wageclaude-sonnet-53/5AI tools reduce data-gathering and drafting time substantially, but licensing, model oversight, and integration with proprietary financial systems keep costs from being dramatically lower than skilled analyst time for high-stakes forecasting.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed financial forecasting tools (Bloomberg, FactSet, specialized ML platforms) reliably produce analyses and economic predictions at scale in production. However, most require human validation and oversight for critical decisions, and accuracy varies by market conditions and asset class.
Technical feasibility todayclaude-sonnet-53/5Deployed products (e.g., Bloomberg GPT, forecasting platforms, LLM-based analytics tools) exist and are used in production, but they still have material error rates and require human validation for high-stakes forecasts.

Evaluate the risks and benefits involved in implementing green building technologies.

60

CI 32–87 · exposure 58 · augmentation 88 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Financial and real estate sectors are rapidly deploying AI for risk analysis, cost modeling, and regulatory compliance. ESG and green building initiatives have high capital stakes, driving accelerated adoption of AI-assisted evaluation tools in production settings.
Sector adoption velocityclaude-sonnet-53/5Financial services broadly show fast AI adoption, but the specific niche of green building technology risk assessment is narrower and less mature in deployment, giving middling velocity.
Augmentation potentialclaude-haiku-4-5-202510015/5AI excels at augmenting risk specialists by rapidly synthesizing diverse data sources, generating scenario models, and identifying relevant regulatory and technical considerations, allowing the human specialist to focus on judgment, strategic implications, and stakeholder communication.
Augmentation potentialclaude-sonnet-54/5AI tools can meaningfully speed up research on technology options, regulatory context, cost-benefit data gathering, and drafting risk summaries, substantially aiding a human specialist's productivity.
Task automatabilityclaude-haiku-4-5-202510015/5AI can comprehensively evaluate green building technologies by analyzing technical specifications, cost-benefit models, regulatory frameworks, and environmental impact data from multiple sources. Systems can synthesize risk-benefit analyses, perform financial modeling, and generate structured reports meeting the ≥50% time-saving threshold with equal quality.
Task automatabilityclaude-sonnet-52/5This requires synthesizing domain-specific technical, financial, and regulatory knowledge with judgment calls about uncertain future costs/benefits; AI can support research and analysis but cannot autonomously produce a reliable end-to-end evaluation at equal quality today.
Adoption barriersclaude-haiku-4-5-202510012/5No licensure requirement mandates human sign-off on green building technology risk evaluations, though organizations may prefer human judgment for strategic decisions and regulatory review. Barriers are primarily organizational preference and risk-appetite governance rather than legal constraints.
Adoption barriersclaude-sonnet-53/5No strict licensing requirement dictates human sign-off, but organizational risk aversion, the need for accountable due diligence, and reliance on credentialed risk assessment create moderate friction against pure automation.
Cost vs. human wageclaude-haiku-4-5-202510015/5AI inference and integration costs for financial modeling, document analysis, and risk assessment are orders of magnitude cheaper than the fully-loaded hourly rate of a specialized financial risk analyst, even with human oversight factored in.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply generate background research and drafts, but given the specialized judgment and verification needed, a human analyst still must do substantial work, keeping the cost differential modest rather than order-of-magnitude.
Technical feasibility todayclaude-haiku-4-5-202510014/5Mature AI products (LLMs, data analytics platforms, financial modeling tools) reliably perform components of this task in production: technology comparison, cost analysis, and regulatory mapping. Minor limitations remain in novel or highly specialized technologies, but core evaluation is deployable at scale today.
Technical feasibility todayclaude-sonnet-52/5No deployed product performs full green-building risk/benefit evaluations reliably in production; general LLM tools can assist with research and drafting but lack validated domain-specific reliability for this niche analysis.

Recommend ways to control or reduce risk.

59

CI 28–90 · exposure 58 · augmentation 88 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510015/5Financial institutions have aggressively adopted algorithmic risk systems, quantitative models, and now AI-assisted dashboards over the past decade. The sector is highly digitized, capital-incentivized to reduce costs, and early in deploying agent-like risk tools.
Sector adoption velocityclaude-sonnet-53/5Financial services is a fast-adopting sector for AI, and risk analytics tools are increasingly used, but true recommendation-generation is still mostly pilot-stage rather than deep production use.
Augmentation potentialclaude-haiku-4-5-202510015/5AI dramatically enhances risk specialists' productivity by generating scenario analyses, stress-test outputs, and control recommendations in seconds rather than hours. Specialists retain judgment and oversight while benefiting from AI's speed and breadth of options.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully help specialists by surfacing risk patterns, summarizing scenarios, and drafting mitigation options, substantially speeding the human's analysis while judgment remains human-led.
Task automatabilityclaude-haiku-4-5-202510015/5AI systems can generate risk mitigation strategies, optimize hedging portfolios, and recommend control measures using financial modeling and scenario analysis. This task fits a structured analytical workflow where AI can produce outputs meeting quality and time-saving thresholds with established frameworks.
Task automatabilityclaude-sonnet-52/5Generating risk mitigation recommendations requires synthesizing firm-specific context, regulatory nuance, and judgment about business tradeoffs that current AI cannot fully replicate end-to-end at equal quality.5x8 time savings possible for drafting but not full task completion.5x8 threshold not met.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8.5x8
Adoption barriersclaude-haiku-4-5-202510012/5While there are industry standards (Basel III, enterprise governance) and some regulatory documentation requirements, no law mandates a licensed specialist personally author risk recommendations. Human sign-off is common practice but not a hard legal barrier to automation.
Adoption barriersclaude-sonnet-54/5Financial risk recommendations often require sign-off from licensed risk officers or compliance functions, and liability for flawed risk advice creates strong organizational and regulatory barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510015/5AI inference for risk recommendation is highly cost-effective—a single query to a deployed model costs pennies, while a specialist analyst hour costs $150–300. Integration and oversight overhead is modest relative to the labor savings.
Cost vs. human wageclaude-sonnet-52/5While AI can cheaply draft generic risk mitigation ideas, the human expert time needed to validate, tailor, and take responsibility for recommendations keeps overall cost close to or above the human-only baseline for high-stakes decisions.
Technical feasibility todayclaude-haiku-4-5-202510014/5Mature financial AI products (Bloomberg, RiskMetrics-type systems, and newer ML-based risk platforms) already recommend risk controls in production across major financial institutions. However, complex enterprise risk decisions often require human validation, keeping it below a perfect 5.
Technical feasibility todayclaude-sonnet-52/5Some risk-analytics platforms and GPT-based tools can draft risk mitigation suggestions, but no deployed product reliably generates authoritative, context-aware risk control recommendations without heavy human validation.

Devise systems or processes to monitor validity of risk assessments.

57

CI 28–87 · exposure 58 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Financial services firms, especially large banks and asset managers, are rapidly deploying automated risk monitoring and model surveillance tools as part of enterprise risk management. This reflects sector-wide digital transformation and regulatory pressure, with measurable production adoption.
Sector adoption velocityclaude-sonnet-53/5Financial services is a fast-adopting sector for AI tools generally, but risk model governance functions adopt more cautiously due to regulatory scrutiny, so uptake here is moderate rather than fast.
Augmentation potentialclaude-haiku-4-5-202510014/5AI-driven monitoring systems significantly augment risk specialists' productivity by surfacing issues, quantifying deviations, and filtering vast datasets for human attention. Specialists retain authority over judgment but operate at much higher throughput and coverage with AI assistance.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by suggesting monitoring metrics, generating code, analyzing historical model performance data, and flagging anomalies, substantially boosting a specialist's productivity while they retain design authority.
Task automatabilityclaude-haiku-4-5-202510015/5Current AI systems can automatically monitor validity of risk assessments through pattern detection, anomaly identification, and deviation flagging across large datasets at scale. This task of systematic monitoring and validity checking is well-suited to deterministic rule engines and ML models, achieving substantial time savings without human intervention.
Task automatabilityclaude-sonnet-52/5Designing monitoring systems requires judgment about institutional risk appetite, regulatory context, and model validation strategy that current AI cannot originate end-to-end, though it can assist with sub-components like drafting metrics or code for dashboards...
Adoption barriersclaude-haiku-4-5-202510012/5While financial regulation (e.g., SR 11-7 model risk governance) requires oversight of risk models, it does not mandate that a human devise the monitoring systems themselves—automation is often encouraged for efficiency and consistency. Implementation may involve some internal governance review, but no legal requirement for human sign-off on the monitoring systems blocks automation.
Adoption barriersclaude-sonnet-54/5Financial risk monitoring is heavily regulated (e.g., SR 11-7, Basel model validation requirements) and typically requires sign-off by qualified risk professionals, creating strong compliance and liability barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510015/5Continuous automated monitoring via ML pipelines and rule engines costs orders of magnitude less than employing humans to manually review and validate risk assessments. Infrastructure scales nearly linearly; human oversight would require proportional headcount.
Cost vs. human wageclaude-sonnet-52/5Because human oversight, domain expertise, and iterative validation are still required, AI mainly reduces some drafting/analysis time rather than replacing the specialist, so cost savings are moderate at best.
Technical feasibility todayclaude-haiku-4-5-202510014/5Production systems exist for risk model monitoring (e.g., model monitoring platforms, backtesting frameworks, anomaly detection in risk systems), though most require some human configuration and interpretation. Deployed products reliably flag deviations and model drift, though complex judgment calls on validity interpretation may still need human review.
Technical feasibility todayclaude-sonnet-52/5There are model risk management and MLOps monitoring tools, but no deployed product autonomously devises full validity-monitoring frameworks for risk assessments in production without heavy human design and configuration.

Evaluate the risks related to green investments, such as renewable energy company stocks.

57

CI 36–77 · exposure 50 · augmentation 88 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510015/5Financial services and asset management firms are rapidly adopting AI-driven ESG and climate risk analysis; major firms (BlackRock, Vanguard, fund managers) now integrate AI risk models into production workflows as core infrastructure.
Sector adoption velocityclaude-sonnet-53/5Financial services broadly show above-average AI adoption, but specialized ESG/green risk assessment is a narrower niche still in pilot phase at many firms, with production deployment uneven across the sector.
Augmentation potentialclaude-haiku-4-5-202510015/5AI substantially augments human analysts by automating data aggregation, stress-testing scenarios, and flagging anomalies, freeing specialists to focus on strategic judgment, novel risks, and stakeholder communication. Productivity gains are measurable and widely adopted.
Augmentation potentialclaude-sonnet-54/5AI substantially aids analysts by rapidly aggregating regulatory changes, climate risk data, and market sentiment, letting human experts focus on judgment-intensive synthesis and decision-making.
Task automatabilityclaude-haiku-4-5-202510014/5AI can automatically gather financial data, analyze sustainability metrics, identify correlations between ESG factors and market performance, and generate risk reports with significant time savings. However, judgment calls on emerging climate policy impact and novel technology viability still benefit from human expertise, preventing a full 5.
Task automatabilityclaude-sonnet-52/5Risk evaluation for green investments requires synthesizing regulatory, market, technology, and financial data with contextual judgment that current AI cannot fully replicate end-to-end; AI can draft parts but a human must integrate and validate the analysis, limiting time savings to well below 50% at equal quality overall.
Adoption barriersclaude-haiku-4-5-202510012/5While senior investment decisions face regulatory oversight and fiduciary duty requirements, green investment risk evaluation itself has no legal mandate that a human must personally perform the analysis. Institutional friction and preference for human sign-off provide modest friction but not hard barriers.
Adoption barriersclaude-sonnet-53/5While not legally mandated to be human-performed, financial risk assessments often require sign-off by licensed professionals (e.g., CFA, risk officers) and carry liability exposure, creating moderate organizational and regulatory friction against full automation.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI-driven risk analysis costs orders of magnitude less than full human analyst review once models are trained and integrated. A single financial analyst's annual loaded cost ($150–250k+) far exceeds the per-task inference and data costs, yielding strong ROI.
Cost vs. human wageclaude-sonnet-53/5AI tools can cheaply process large volumes of data (filings, news, climate data) at a fraction of analyst cost, but the need for expert review and integration keeps overall cost roughly comparable to a human-led process once quality assurance is factored in.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed systems (Bloomberg terminals, Refinitiv, MSCI ESG ratings) already perform green investment risk analysis at scale in production. Outputs include ESG scores, carbon footprint analysis, and regulatory risk flags, though interpretation and edge-case handling still require human oversight.
Technical feasibility todayclaude-sonnet-52/5There are analytics and NLP tools (e.g., ESG data platforms, sentiment analysis) that assist in gathering information, but no deployed product independently produces a reliable, decision-grade risk evaluation of green investments without heavy human oversight.

Review or draft risk disclosures for offer documents.

57

CI 40–74 · exposure 66 · augmentation 88 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Finance and legal departments have rapidly adopted AI-assisted document review and drafting tools, especially in large firms and investment banks; disclosure automation is already commonplace in high-volume M&A and equity issuance workflows.
Sector adoption velocityclaude-sonnet-53/5Financial services broadly show middling-to-fast AI adoption for drafting and compliance support, but risk disclosure drafting specifically remains cautious due to legal exposure, with pilots more common than full production deployment.
Augmentation potentialclaude-haiku-4-5-202510015/5AI dramatically enhances productivity by auto-generating first drafts, flagging missing disclosures against regulatory checklists, and accelerating review cycles, allowing specialists to focus on judgment and materiality assessment rather than boilerplate assembly.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully speed up drafting, flag missing risk factors, and check consistency against prior disclosures or regulatory templates, substantially aiding specialists while they retain final judgment and sign-off.
Task automatabilityclaude-haiku-4-5-202510015/5AI systems can generate well-structured draft risk disclosures from templates, regulatory frameworks, and transaction data with minimal human intervention, easily meeting the 50% time-saving threshold. Legal tech and LLM-based document assembly tools already handle large portions of disclosure drafting for M&A and securities offerings.
Task automatabilityclaude-sonnet-53/5AI can draft initial risk disclosure language and identify boilerplate risk categories, but final review requires nuanced legal/regulatory judgment and firm-specific risk assessment that current systems cannot fully replicate end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Financial regulations and securities law (SEC, FINRA, prospectus rules) impose legal liability for disclosure accuracy and completeness, effectively requiring a licensed financial professional or legal specialist to review and approve final disclosures; automated drafts do not eliminate the human gatekeeper.
Adoption barriersclaude-sonnet-54/5Offer documents are subject to securities regulations requiring qualified professionals to review and attest to disclosure accuracy, creating strong liability and regulatory barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI-assisted drafting and review costs a fraction of senior specialist labor; inference and document generation are cheap compared to human time at $150–300/hour billing rates, achieving meaningful cost displacement.
Cost vs. human wageclaude-sonnet-53/5AI drafting assistance can reduce time spent on initial drafts significantly, but the need for expert legal/compliance review and liability oversight keeps all-in costs closer to comparable with human specialists rather than order-of-magnitude cheaper.
Technical feasibility todayclaude-haiku-4-5-202510014/5Multiple deployed products (Kira, LawGeex, specialized disclosure platforms) reliably draft and review risk disclosure sections in real organizations, though final sign-off and judgment on materiality still typically require expert human review due to liability sensitivity.
Technical feasibility todayclaude-sonnet-52/5Some legal-drafting and compliance AI tools exist that assist with disclosure language, but production systems reliably handling full risk disclosure drafting for offer documents in regulated financial contexts are narrow and require heavy human review.

Prepare plans of action for investment, using financial analyses.

55

CI 28–82 · exposure 58 · augmentation 88 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Financial services sectors show rapid AI adoption for analytics and planning support, with robo-advisory platforms, algorithmic trading, and AI-assisted portfolio management already in widespread institutional and consumer use.
Sector adoption velocityclaude-sonnet-53/5Finance is a fast-adopting sector for AI analytics tools, but actual autonomous plan generation without human sign-off is still limited to pilots and augmented workflows.
Augmentation potentialclaude-haiku-4-5-202510015/5AI assistants substantially boost financial specialist productivity by automating data gathering, scenario analysis, and initial plan drafting, allowing humans to focus on client strategy and risk judgment.
Augmentation potentialclaude-sonnet-54/5AI substantially speeds up data gathering, scenario modeling, and drafting for financial analyses, letting specialists focus on judgment and strategic decision-making.
Task automatabilityclaude-haiku-4-5-202510015/5Current AI systems can perform financial analysis, scenario modeling, and generate structured investment plans at scale with minimal human intervention, meeting the ≥50% time-saving threshold for significant portions of this task.
Task automatabilityclaude-sonnet-52/5AI can generate draft analyses and scenario outputs, but synthesizing them into a defensible investment action plan requires judgment, risk appetite context, and accountability that current systems cannot fully replicate end-to-end.
Adoption barriersclaude-haiku-4-5-202510013/5Regulatory oversight, client fiduciary responsibility, and risk liability create moderate friction; a licensed financial advisor typically must review and authorize final investment plans, preventing full substitution despite technical capability.
Adoption barriersclaude-sonnet-54/5Financial advice and investment planning are often subject to regulatory licensing (e.g., fiduciary duty, suitability rules) and liability concerns, requiring qualified professionals to approve or sign off on final plans.
Cost vs. human wageclaude-haiku-4-5-202510015/5AI-driven financial analysis and plan generation costs orders of magnitude less than specialist human labor, with inference and integration costs well below loaded wages for financial analysts.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply produce quantitative analysis components, but the human oversight, validation, and compliance review needed for actionable investment plans keeps blended costs closer to human-comparable.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed products (Bloomberg terminals, robo-advisors, risk analytics platforms with AI modules) reliably perform financial analysis and plan generation in production, though strategic judgment and regulatory sign-off typically remain human responsibilities.
Technical feasibility todayclaude-sonnet-52/5Products like AI-assisted portfolio analytics and robo-advisory tools exist, but reliable production-grade generation of full investment action plans for professional risk management remains narrow and closely supervised.

Identify key risks and mitigating factors of potential investments, such as asset types and values, legal and ownership structures, professional reputations, customer bases, or industry segments.

55

CI 45–65 · exposure 58 · augmentation 100 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Financial services and asset management firms are actively deploying AI-driven risk assessment, ESG scoring, and due-diligence automation in production systems. Adoption is rapid among larger institutions and grows steadily in wealth management and private equity, reflecting digital infrastructure maturity and competitive pressure in the sector.
Sector adoption velocityclaude-sonnet-53/5Financial services is a fast-adopting sector for AI copilots and analytics, but risk assessment specifically still relies heavily on human expert judgment, so deployment is more pilot/augmentation than full replacement.
Augmentation potentialclaude-haiku-4-5-202510015/5AI dramatically augments specialist productivity by rapidly screening large document sets, cross-referencing regulatory databases, and flagging anomalies in ownership and financial structures. Specialists retain control over final judgment, but AI-assisted analysis materially accelerates and depersonalizes the evidence-gathering phase of investment risk assessment.
Augmentation potentialclaude-sonnet-55/5AI dramatically accelerates data gathering, pattern detection, and drafting of risk summaries, letting specialists focus judgment on the highest-value assessment while AI handles broad information synthesis.
Task automatabilityclaude-haiku-4-5-202510014/5Current AI systems can analyze structured financial data, legal documents, and industry reports to identify standard risk categories and mitigating factors with significant time savings. However, the task requires nuanced judgment about emerging or non-standard risks and reputational factors that may involve qualitative research beyond automated extraction, making consistent 50% time savings achievable but not universal across all investment types.
Task automatabilityclaude-sonnet-53/5AI can synthesize public data, financial statements, and industry reports to surface risk factors, but nuanced judgment on legal structures, reputational risk, and materiality requires human validation, limiting full end-to-end automation.
Adoption barriersclaude-haiku-4-5-202510013/5Regulatory frameworks (e.g., securities law, lending standards) and organizational risk governance often require a licensed professional to sign off on investment decisions, creating a hard human-in-the-loop requirement. However, AI can augment the analysis phase without replacing the licensed decision-maker, limiting but not eliminating substitution.
Adoption barriersclaude-sonnet-54/5Investment risk assessments often carry regulatory and fiduciary responsibilities, requiring qualified professionals to validate conclusions, creating meaningful liability and compliance barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI-driven risk screening and document analysis cost a fraction of specialist time (per-investment inference costs are minimal once integrated), easily yielding 5–10× cost advantage when accounting for scale and reduced manual due diligence overhead. Residual human oversight is still required but represents partial cost savings relative to fully manual analysis.
Cost vs. human wageclaude-sonnet-53/5AI can cut research time significantly, but the need for licensed analysts to verify and sign off on risk assessments keeps blended cost closer to comparable rather than order-of-magnitude cheaper.
Technical feasibility todayclaude-haiku-4-5-202510013/5Multiple AI-powered risk analysis platforms and LLM-based document analysis tools exist and perform parts of this task reliably in production (financial data extraction, regulatory screening, industry segment classification). However, material gaps remain in assessing professional reputation credibility and ownership structure complexity, and false negatives on novel risks remain concerning enough that human specialist review is standard practice.
Technical feasibility todayclaude-sonnet-53/5Deployed tools (e.g., risk analytics platforms with LLM features, due-diligence software) assist in flagging risks, but reliable comprehensive risk identification across asset types and legal structures still shows material gaps and requires analyst review.

Interpret data on price, yield, stability, future investment-risk trends, economic influences, and other factors affecting investment programs.

49

CI 42–56 · exposure 45 · augmentation 100 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Financial services and asset management firms are rapidly adopting AI-driven risk analytics, quantitative models, and data interpretation platforms in production. Pilot-to-deployment timelines are shortening, and competitive pressure is driving adoption faster than in most other sectors.
Sector adoption velocityclaude-sonnet-54/5Finance is a fast-adopting sector for AI-driven analytics, with widespread deployment of quantitative and NLP tools in risk and investment functions.
Augmentation potentialclaude-haiku-4-5-202510015/5AI substantially augments specialist productivity by automating data aggregation, computing risk metrics, flagging anomalies, and generating scenario analyses—freeing the human to focus on judgment, stress-testing assumptions, and strategic recommendation. This is a textbook case of high-value augmentation while the specialist remains in control.
Augmentation potentialclaude-sonnet-55/5AI significantly enhances human capability in this task by rapidly synthesizing large datasets, market signals, and trend analysis for the specialist to interpret and validate.
Task automatabilityclaude-haiku-4-5-202510013/5AI can extract and summarize data on price, yield, economic indicators, and compute basic risk metrics (volatility, correlation) at scale, but interpretation requiring contextual judgment, tail-risk assessment, and integration of nonlinear interdependencies between factors remains challenging. A human specialist would still need to validate conclusions and make final risk decisions.
Task automatabilityclaude-sonnet-53/5AI can process quantitative data, generate summaries, and flag trends, but synthesizing this into judgment-based investment-risk interpretation requiring accountability and contextual nuance still needs significant human oversight.'.'}
Adoption barriersclaude-haiku-4-5-202510014/5Significant regulatory and liability barriers exist: securities regulators require documented human oversight of investment recommendations, fiduciary duty implies accountability that is hard to delegate fully to AI, and errors in risk interpretation carry material legal and financial consequences that firms are reluctant to outsource entirely.
Adoption barriersclaude-sonnet-53/5No explicit licensing requirement to interpret data itself, but firms have compliance, fiduciary duty, and liability concerns that create moderate friction against fully automating investment risk judgments.
Cost vs. human wageclaude-haiku-4-5-202510014/5Modern risk-analytics platforms and LLM-powered data synthesis cost a fraction of a mid-level financial analyst's loaded salary, especially once amortized across portfolios. Integration overhead is modest in digitized financial firms, making AI cost substantially cheaper per interpretation task.
Cost vs. human wageclaude-sonnet-53/5AI tools reduce data-processing time meaningfully but human analysts still needed for validation and judgment, so all-in cost is roughly comparable to human-only workflows with AI assistance layered in.
Technical feasibility todayclaude-haiku-4-5-202510013/5Deployed AI systems (Bloomberg terminals, risk platforms, ML-driven analytics) successfully flag pricing anomalies, compute standard risk metrics, and surface relevant economic data, but production systems rarely operate end-to-end without expert review. Material error rates on non-standard scenarios and extreme events limit full autonomy.
Technical feasibility todayclaude-sonnet-52/5Products exist that provide analytics and risk dashboards, but full autonomous interpretation with reliable judgment across varied economic contexts is not yet a mature production capability.

Determine potential environmental impacts of new products or processes on long-term growth and profitability.

47

CI 32–62 · exposure 45 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Large financial institutions have begun integrating AI-driven ESG and climate-risk tools into workflows, but adoption remains selective and often limited to pilot phases or reporting support rather than core decision-making at scale across the sector.
Sector adoption velocityclaude-sonnet-53/5Financial services broadly adopts AI quickly, but ESG/environmental risk modeling is a specialized niche with slower, more cautious uptake due to data quality and regulatory uncertainty.
Augmentation potentialclaude-haiku-4-5-202510014/5AI substantially assists risk specialists by rapidly synthesizing environmental data, modeling scenarios, and identifying exposure vectors, allowing analysts to focus on complex judgment calls and stakeholder communication rather than data compilation.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by rapidly aggregating environmental data, regulatory changes, climate scenarios, and market research, significantly speeding up the analyst's information-gathering and drafting process.
Task automatabilityclaude-haiku-4-5-202510014/5AI can systematically analyze environmental impacts using regulatory frameworks, ESG datasets, and lifecycle assessment models to flag risks and project long-term financial effects with minimal human setup. However, novel or unprecedented environmental scenarios may still require human judgment, preventing a full 5.
Task automatabilityclaude-sonnet-52/5This requires synthesizing environmental science, regulatory trends, market forecasts, and company-specific data into forward-looking judgment calls; AI can support pieces but cannot autonomously produce reliable, defensible long-term impact assessments end-to-end.
Adoption barriersclaude-haiku-4-5-202510013/5Regulatory frameworks (SEC climate disclosure, TCFD) and fiduciary duty standards require some form of documented analysis and expert sign-off, creating moderate friction. However, no license explicitly forbids AI-assisted or AI-driven environmental risk assessment, leaving space for automation.
Adoption barriersclaude-sonnet-53/5No licensing requirement mandates a human specifically, but internal governance, fiduciary responsibility, and reputational/liability risk around environmental and financial projections create meaningful oversight friction.
Cost vs. human wageclaude-haiku-4-5-202510014/5Environmental risk analysis via AI tools (data + inference) costs orders of magnitude less than hiring specialized environmental consultants or conducting bespoke impact studies, though human oversight remains necessary.
Cost vs. human wageclaude-sonnet-52/5AI tools can cut research time but the analysis still requires substantial human oversight, domain expertise, and validation, so total cost savings versus a skilled analyst are modest rather than order-of-magnitude.
Technical feasibility todayclaude-haiku-4-5-202510013/5ESG analysis and environmental impact assessment tools exist in production (Bloomberg, Refinitiv, specialized climate-risk platforms), but they often require significant expert interpretation and typically cover narrow scopes or rely on incomplete forward-looking data, limiting reliability on novel products.
Technical feasibility todayclaude-sonnet-52/5Some ESG analytics and scenario-modeling tools exist, but they are narrow, require heavy customization, and are not demonstrated to reliably replace analyst judgment on this specific forward-looking synthesis task in production.

Produce reports or presentations that outline findings, explain risk positions, or recommend changes.

47

CI 41–53 · exposure 50 · augmentation 88 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Financial services are early adopters of AI tooling; major institutions are piloting AI-assisted report generation and are moving toward production use in lower-stakes reporting, though higher-stakes risk assessments remain heavily human-led.
Sector adoption velocityclaude-sonnet-53/5Financial services is a fast-digitizing sector with growing pilot use of generative AI for reporting, but risk/compliance functions are more cautious, so production deployment lags leading-edge fintech use cases.
Augmentation potentialclaude-haiku-4-5-202510015/5AI is actively used to augment risk analysts—auto-generating boilerplate sections, pulling and formatting data, suggesting visualizations, and drafting initial framing—allowing experts to focus on judgment, validation, and strategic insight rather than manual composition.
Augmentation potentialclaude-sonnet-54/5AI substantially speeds up drafting, summarizing data trends, and formatting presentations, letting risk specialists focus on judgment and validation rather than manual writing.
Task automatabilityclaude-haiku-4-5-202510013/5AI can draft substantial portions of risk reports (data summaries, standard risk metrics, basic explanations) with good efficiency, but the synthesis of complex findings, strategic recommendations, and context-specific positioning typically requires human expert judgment and carries legal/compliance weight that demands human review and sign-off.
Task automatabilityclaude-sonnet-53/5AI can draft report narratives, summarize data, and generate slides from structured inputs, but synthesizing firm-specific risk judgment and ensuring accuracy still requires substantial human review, limiting full end-to-end automation.
Adoption barriersclaude-haiku-4-5-202510014/5Financial risk reports often feed into regulatory filings, risk governance bodies, or board-level decisions where liability, audit trails, and legal sign-off by licensed professionals (CFA, FRM holders, or compliance-approved signatories) create hard barriers to fully autonomous automation.
Adoption barriersclaude-sonnet-53/5No strict licensing requirement to write these reports, but regulatory scrutiny, internal sign-off chains, and liability for risk misstatements create meaningful institutional friction.
Cost vs. human wageclaude-haiku-4-5-202510012/5While AI drafting reduces composition time, the critical review and refinement by skilled financial risk analysts remains expensive; the all-in cost (inference, integration, senior oversight) approaches or sometimes exceeds the cost of a more junior analyst drafting from scratch, especially given compliance requirements.
Cost vs. human wageclaude-sonnet-53/5AI drafting tools are cheap per use, but the oversight, data integration, and validation needed for risk reporting narrows the cost advantage relative to a skilled analyst's time saved.
Technical feasibility todayclaude-haiku-4-5-202510013/5Production systems exist (LLMs, BI tools with AI) that can generate initial report drafts and presentation outlines, but material error rates in quantitative accuracy, nuance in risk interpretation, and alignment with regulatory/organizational requirements mean they are used primarily for acceleration, not autonomous output.
Technical feasibility todayclaude-sonnet-53/5Products like Copilot, ChatGPT enterprise, and BI tools with LLM summarization are used in production for drafting reports, but reliability for nuanced risk interpretation and compliance-grade output remains inconsistent.

Develop or implement risk-assessment models or methodologies.

44

CI 28–60 · exposure 45 · augmentation 88 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Financial services firms have rapidly adopted AI for model development, with major banks and hedge funds deploying ML-based risk tools in production. Adoption is concentrated in large, digitally mature institutions with high investment in fintech and significant regulatory compliance infrastructure.
Sector adoption velocityclaude-sonnet-53/5Financial services is a fast-adopting sector for AI tools generally, but for core risk model development specifically, adoption remains at the pilot/augmentation stage due to governance and regulatory constraints.
Augmentation potentialclaude-haiku-4-5-202510015/5AI dramatically assists risk specialists through automated feature engineering, rapid prototyping, scenario simulation, and backtesting, enabling faster iteration and exploration of model families. Specialists remain central to interpretation, governance, and strategic alignment, making this a high-augmentation scenario.
Augmentation potentialclaude-sonnet-54/5AI significantly aids risk specialists in tasks like data analysis, scenario generation, code drafting, and literature synthesis, meaningfully boosting productivity while humans retain model ownership and judgment.
Task automatabilityclaude-haiku-4-5-202510014/5AI can automate substantial portions of model development (algorithm selection, hyperparameter tuning, backtesting) and implementation workflows, achieving significant time savings. However, final validation, stakeholder review, and integration into legacy systems typically require human judgment and coordination, preventing full end-to-end automation at the ≥50% threshold for independent specialists.
Task automatabilityclaude-sonnet-52/5Developing risk models requires domain expertise, judgment about assumptions, regulatory context, and validation against business realities that current AI cannot fully replace end-to-end, though AI can assist with coding, statistical analysis, and literature review.
Adoption barriersclaude-haiku-4-5-202510014/5Regulatory requirements (Basel III, Dodd-Frank, MiFID II) often mandate that humans validate, govern, and sign off on risk models; model interpretability and explainability standards create friction for black-box AI implementations. Liability asymmetry (model failures can trigger severe penalties) further protects human oversight roles.
Adoption barriersclaude-sonnet-54/5Financial risk models are often subject to regulatory scrutiny (e.g., model risk management frameworks like SR 11-7) requiring qualified human sign-off, validation, and accountability, creating strong barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI-powered model development and backtesting reduce infrastructure and analyst-hours substantially, making incremental automation cheaper than hiring additional specialists. However, oversight, validation, and integration overhead prevent the order-of-magnitude cost advantage seen in lower-complexity tasks.
Cost vs. human wageclaude-sonnet-52/5AI can reduce time spent on coding and data exploration, but the overall cost includes expensive expert oversight, validation, and regulatory review, keeping costs closer to comparable rather than order-of-magnitude cheaper.
Technical feasibility todayclaude-haiku-4-5-202510013/5Mature ML platforms (AutoML, model-building frameworks) and risk analytics vendors exist and are deployed in production; however, they are often applied to narrow domains (credit scoring, market risk) and require substantial customization and domain expertise to adapt to novel risk scenarios or regulatory requirements.
Technical feasibility todayclaude-sonnet-52/5Some fintech and quant tools use AI to assist in model development (feature selection, backtesting), but no deployed product autonomously develops and validates full risk-assessment methodologies at production quality without heavy human oversight.

Maintain input or data quality of risk management systems.

43

CI 36–50 · exposure 38 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Financial services have piloted data quality automation and ML-based monitoring widely, but production deployment remains mixed; most firms still rely on hybrid manual-automated workflows rather than fully autonomous systems, reflecting moderate adoption momentum.
Sector adoption velocityclaude-sonnet-53/5Financial services broadly adopt AI/ML for data quality and monitoring, but risk management specifically tends to be conservative and cautious due to regulatory scrutiny, placing it in the middle of adoption curves.
Augmentation potentialclaude-haiku-4-5-202510014/5AI dashboards, automated anomaly alerts, and suggested corrections significantly boost specialist productivity by filtering noise and prioritizing high-risk issues, allowing human experts to focus on interpretation and remediation rather than routine data scanning.
Augmentation potentialclaude-sonnet-54/5AI-based anomaly detection, automated reconciliation, and data profiling tools significantly help specialists find quality issues faster and prioritize remediation while humans retain final judgment and sign-off.
Task automatabilityclaude-haiku-4-5-202510013/5AI can automate significant portions of data quality checks (validation rules, anomaly detection, pattern matching) and flag inconsistencies, but domain expertise is needed to interpret edge cases and determine appropriate corrective actions, making full end-to-end automation with quality parity difficult without human oversight.
Task automatabilityclaude-sonnet-52/5Data quality maintenance requires ongoing judgment about anomalies, source reliability, and system-specific context that current AI can assist with but not fully own end-to-end.》 Partial automation is feasible for validation checks, but full task substitution is not yet reliable.
Adoption barriersclaude-haiku-4-5-202510013/5Risk management systems often operate under regulatory scrutiny (Basel, Dodd-Frank, etc.) and organizational governance requires human sign-off on data quality decisions; however, these are process/oversight barriers rather than hard legal bans on automation, leaving room for delegated monitoring.
Adoption barriersclaude-sonnet-53/5Risk management data quality is often subject to regulatory expectations (e.g., BCBS 239, model risk governance) requiring documented human accountability, creating moderate barriers even though the underlying data checks aren't licensed activities.
Cost vs. human wageclaude-haiku-4-5-202510013/5Automated data quality monitoring is operationally cheaper than manual inspection at scale, but integration, tuning, and ongoing oversight by specialists still represent meaningful costs comparable to hiring mid-level data quality personnel for medium-sized organizations.
Cost vs. human wageclaude-sonnet-53/5Automated validation rules and anomaly detection reduce manual review costs somewhat, but the human oversight, exception handling, and governance required keep costs roughly comparable to a human specialist doing full quality assurance.
Technical feasibility todayclaude-haiku-4-5-202510013/5Data quality and monitoring tools with ML components exist in production (e.g., data profiling, outlier detection platforms), but they typically require substantial configuration and human validation; no mature system fully owns data quality maintenance for complex risk management systems without expert review.
Technical feasibility todayclaude-sonnet-52/5Data quality/validation tools and anomaly detection products exist and are used in production, but comprehensive data quality management for risk systems still requires human oversight and remediation, so deployed products cover only part of the task.

Analyze areas of potential risk to the assets, earning capacity, or success of organizations.

40

CI 28–53 · exposure 38 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Financial services and enterprise risk management are digitized sectors with active AI adoption; major banks and insurers deploy machine learning risk models in production. However, adoption remains concentrated in high-volume, standardized risk categories (credit, market, operational compliance) rather than holistic organizational risk assessment.
Sector adoption velocityclaude-sonnet-53/5Financial services is a fast-adopting sector for AI generally, but adoption specifically for holistic enterprise risk analysis remains at pilot/tool-assisted stage rather than full production autonomy.
Augmentation potentialclaude-haiku-4-5-202510014/5AI assistants demonstrably augment risk specialists by automating data collection, running scenario analyses, surfacing anomalies, and generating preliminary reports, allowing experts to focus on judgment and interpretation. Risk dashboards and predictive models are widely used to enhance analyst productivity while maintaining human oversight.
Augmentation potentialclaude-sonnet-54/5AI substantially aids risk specialists via data analysis, pattern detection, scenario simulation, and report drafting, meaningfully increasing productivity while humans retain interpretive and decision-making control.
Task automatabilityclaude-haiku-4-5-202510013/5AI can automate parts of risk analysis—data aggregation, pattern detection in historical risk events, and routine scenario modeling—but the task requires contextual judgment about organizational strategy, competitive dynamics, and tail risks that remain difficult for current systems. A human expert would still be needed to interpret findings and prioritize risks.
Task automatabilityclaude-sonnet-52/5Risk analysis requires synthesizing quantitative data, organizational context, and judgment about uncertain future events; AI can support pieces (data aggregation, scenario modeling) but cannot independently perform the full analytical and judgment-laden task at equal quality yet.'
Adoption barriersclaude-haiku-4-5-202510014/5Financial risk analysis is heavily regulated (Basel III, Dodd-Frank, SOX compliance) and often requires sign-off by licensed professionals (CFOs, risk officers, auditors). Liability for missed or miscalculated risks creates strong error-cost asymmetry, and regulatory frameworks explicitly mandate human accountability for risk governance.
Adoption barriersclaude-sonnet-53/5No strict licensing mandate for this specific analytical task, but regulatory expectations (e.g., SOX, Basel, internal audit sign-off) and liability concerns create moderate friction against full automation.
Cost vs. human wageclaude-haiku-4-5-202510013/5AI-powered risk platforms have moderate to high implementation costs (data integration, model tuning, validation) and ongoing operational overhead, roughly comparable to mid-level analyst wages when fully amortized. Some data-intensive routine tasks are now cheaper via AI, but comprehensive organizational risk analysis still requires substantial human oversight.
Cost vs. human wageclaude-sonnet-52/5AI tools can cheaply process data and flag risk indicators, but the full task still requires expensive human expert review, oversight, and judgment, keeping blended costs closer to human-comparable levels.
Technical feasibility todayclaude-haiku-4-5-202510013/5Risk analytics tools exist in production (fraud detection, credit scoring, market risk systems), but they typically handle narrow, well-defined risk categories with material error rates. Broad organizational risk analysis across assets, earnings capacity, and strategic success remains partially manual; no single deployed product reliably performs the full scope.
Technical feasibility todayclaude-sonnet-52/5Deployed risk analytics tools exist (e.g., credit risk models, fraud detection) but comprehensive enterprise risk analysis spanning assets, earnings, and strategic risk is still largely human-led with AI as a supporting input.

Develop contingency plans to deal with emergencies.

33

CI 25–41 · exposure 33 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Financial services adoption of AI for contingency planning remains limited to pilot programs and analytical support; most institutions rely on expert-driven planning with human oversight, reflecting both regulatory conservatism and the high stakes of emergency response accuracy.
Sector adoption velocityclaude-sonnet-53/5Financial services is a fast-adopting sector for AI generally, but contingency/risk planning specifically remains a slower-adopting, high-stakes niche with mostly pilot-stage tool use.
Augmentation potentialclaude-haiku-4-5-202510014/5AI substantially assists by automating scenario generation, regulatory requirement mapping, historical precedent analysis, and draft documentation, allowing risk specialists to focus on strategic decision-making and stakeholder alignment rather than manual compilation and routine scenario synthesis.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by generating scenario lists, drafting plan structures, summarizing historical incidents, and stress-testing assumptions, significantly speeding up the specialist's workflow while they retain final judgment.
Task automatabilityclaude-haiku-4-5-202510013/5AI can generate and structure contingency plans with substantial time savings (data aggregation, scenario modeling, draft documentation), but final plans require human judgment on organizational priorities, risk tolerance, and context-specific decision logic that current systems cannot reliably assess end-to-end.
Task automatabilityclaude-sonnet-52/5Contingency planning requires judgment about organizational priorities, risk tolerance, and stakeholder coordination that AI cannot fully replicate; AI can draft templates but cannot own the strategic decisions or validate real-world feasibility.
Adoption barriersclaude-haiku-4-5-202510014/5Contingency plans for financial emergencies typically require sign-off from senior risk officers, boards, or compliance/regulatory bodies; liability exposure is asymmetric (failure of a defective plan is costly), and many institutions prefer human accountability and human judgment on crisis response.
Adoption barriersclaude-sonnet-53/5No strict licensing requirement mandates a human sign-off, but liability exposure and regulatory expectations around risk management (e.g., financial regulators) create strong incentives for human accountability in contingency planning.
Cost vs. human wageclaude-haiku-4-5-202510012/5Integration of AI planning tools requires significant domain-specific customization, human expert review, and governance overhead; the loaded cost of a financial risk specialist's time spent validating and refining AI outputs often approaches or exceeds the cost of human-directed planning from scratch.
Cost vs. human wageclaude-sonnet-52/5While AI drafting is cheap, the human risk specialist time for validation, stakeholder alignment, and scenario testing remains substantial, keeping overall costs close to human-only baseline.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can assist with drafting and analysis, no mature production system reliably generates validated contingency plans for complex emergencies; existing tools focus on parts (scenario simulation, data synthesis) rather than complete, auditable end-to-end planning that satisfies regulatory and organizational governance.
Technical feasibility todayclaude-sonnet-52/5AI tools can generate draft contingency plan text and scenario checklists, but no deployed product reliably produces validated, organization-specific emergency plans without heavy human rework.

Analyze new legislation to determine impact on risk exposure.

32

CI 28–36 · exposure 25 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Financial services are digitized and early AI adopters, but regulatory risk analysis remains cautious; firms are piloting AI-assisted compliance monitoring and document review, but few have moved to full autonomous impact assessment in production.
Sector adoption velocityclaude-sonnet-53/5Financial services is a fast-adopting sector overall, but risk/compliance functions tend to be more cautious due to regulatory scrutiny, so adoption here lags front-office analytics.
Augmentation potentialclaude-haiku-4-5-202510014/5AI excels at rapidly surfacing relevant legislative language, flagging keyword matches, and generating initial summaries that specialists then review and refine; this materially accelerates the triage and scoping phase of impact analysis while the human retains judgment on materiality.
Augmentation potentialclaude-sonnet-54/5AI tools can significantly speed up legislative review, extraction of key provisions, and drafting of impact summaries, meaningfully boosting analyst productivity while humans retain final judgment.
Task automatabilityclaude-haiku-4-5-202510012/5AI can extract and summarize legislative text and flag potential regulatory changes, but determining *impact* on risk exposure requires domain expertise, institutional context, and judgment about complex financial interactions that current systems cannot reliably perform end-to-end without significant expert oversight.
Task automatabilityclaude-sonnet-52/5AI can summarize legislation and flag relevant clauses, but determining actual risk exposure impact requires domain judgment, organizational context, and interpretive reasoning that current systems cannot fully replace end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Risk analysis underlying capital adequacy and regulatory compliance often has implicit or explicit sign-off requirements by qualified personnel; liability for incorrect impact assessment creates asymmetric error costs that deter full automation.
Adoption barriersclaude-sonnet-53/5No strict licensing requirement for this analytical task, but liability for misjudging regulatory risk exposure creates strong incentive for human sign-off, and compliance functions often have internal governance requirements.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI tools for document review and summarization are relatively inexpensive, but the requirement for specialist oversight and validation of risk impact assessments means total cost per completed analysis remains comparable to human specialists.
Cost vs. human wageclaude-sonnet-53/5AI can cheaply do first-pass reading and summarization, but the analytical judgment portion still requires expensive expert review, keeping overall cost roughly comparable to human-only workflows once oversight is included.
Technical feasibility todayclaude-haiku-4-5-202510012/5While NLP products can identify regulatory changes in legislative documents, no deployed system reliably translates legislation into quantified risk exposure impacts; financial institutions still require specialist review of AI outputs, limiting production automation.
Technical feasibility todayclaude-sonnet-52/5Legal/regulatory summarization tools exist but reliably mapping new legislation to firm-specific risk exposure is not yet a mature, widely deployed production capability; error rates on nuanced interpretation remain material.

Document, and ensure communication of, key risks.

31

CI 25–36 · exposure 25 · augmentation 63 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5While financial services adopt AI for data processing, the core task of identifying and communicating key risks remains human-driven in production. Risk management culture emphasizes human expertise and accountability, limiting fast displacement. Adoption is slow to measured in this specific task despite sector digitization.
Sector adoption velocityclaude-sonnet-53/5Financial services is a fast-adopting sector for AI tools in analytics and reporting, but risk communication specifically still relies heavily on human judgment and pilots are more common than full production deployment.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist meaningfully by surfacing risk signals from large datasets, generating draft documentation, and flagging potential communication gaps, thereby raising the efficiency and thoroughness of specialists. However, the human must retain judgment and final authority, so augmentation is useful but not transformative.
Augmentation potentialclaude-sonnet-54/5AI tools significantly help draft, summarize, and structure risk documentation and can flag anomalies, meaningfully boosting the productivity of human risk specialists while they retain interpretive and communication responsibility.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can help identify, extract, and draft risk documentation from data sources, the task requires judgment about which risks are 'key,' organizational context, and nuanced communication tailored to different stakeholders. Current systems cannot reliably make these prioritization and contextualization decisions without significant human oversight, falling short of the 50% time-saving threshold for end-to-end automation.
Task automatabilityclaude-sonnet-52/5AI can draft risk documentation and summarize findings, but ensuring effective communication and judgment about which risks are 'key' requires organizational context and accountability that current AI cannot fully replace end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Financial risk communication is heavily regulated under frameworks like Dodd-Frank, Basel III, and internal governance policies. Liability for inaccurate risk disclosure is material; human accountability and sign-off are often legally or organizationally required. Risk specialists typically must review and certify risk documentation, creating strong barriers to full automation.
Adoption barriersclaude-sonnet-53/5While not always legally mandated to be human-only, risk documentation in regulated industries (banking, insurance) often requires sign-off by accountable risk officers, creating moderate organizational and compliance friction.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI inference and integration costs for risk analysis remain moderate, but the high human oversight required—financial risk specialists must validate AI outputs, ensure accuracy, and take responsibility for communication—means total cost per reliable task completion is still comparable to or exceeds unaugmented human labor.
Cost vs. human wageclaude-sonnet-53/5AI can cheaply draft documentation, lowering costs for the writing portion, but human review, validation, and stakeholder communication remain necessary, keeping overall cost roughly comparable to human-only workflows.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI tools can assist with risk data aggregation and document generation, but no mature product reliably performs the full task of identifying, prioritizing, and communicating key risks with the judgment and accountability required in financial contexts. Deployed systems exist for narrow subtasks (e.g., regulatory document drafting) but not for the holistic task.
Technical feasibility todayclaude-sonnet-52/5Products exist for drafting reports and summarizing data (e.g., LLM-based report generators), but no deployed system reliably identifies and communicates key risks across an organization without heavy human curation.

Provide statistical modeling advice to other departments.

31

CI 25–36 · exposure 25 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5While finance sectors digitize rapidly, autonomous statistical advisory remains pilot-stage; most organizations treat AI-generated models as drafts requiring expert review rather than as standalone decision support. Adoption is slower than for routine transactional automation.
Sector adoption velocityclaude-sonnet-53/5Financial services broadly show above-average AI adoption, but specialized advisory functions involving cross-department consultation and judgment lag behind more transactional applications.
Augmentation potentialclaude-haiku-4-5-202510014/5AI excels at accelerating statistical modeling tasks—rapid model generation, code suggestions, literature review summaries—allowing human risk specialists to focus on interpreting results and advising stakeholders. LLM and AutoML tools substantially boost productivity when specialists remain in the loop.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully speed up literature review, code generation, exploratory data analysis, and drafting of technical explanations, significantly boosting the productivity of a human specialist providing this advice.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can generate statistical models and code, providing context-aware advice to other departments requires understanding their specific business needs, constraints, and risk tolerance—elements that typically demand human judgment and conversation. Current AI can draft model recommendations but cannot reliably capture the nuance of cross-departmental consultation at scale.
Task automatabilityclaude-sonnet-52/5Providing bespoke statistical modeling advice requires understanding organizational context, stakeholder needs, and judgment calls that current AI cannot fully replicate end-to-end, though it can assist with parts of the analysis.
Adoption barriersclaude-haiku-4-5-202510014/5Financial risk advice carries liability exposure; regulatory frameworks (banking, insurance) often require qualified professionals to sign off on material risk assessments. Organizational risk committees typically demand human accountability and expert sign-off on statistical advice affecting capital or compliance decisions.
Adoption barriersclaude-sonnet-53/5No strict licensing requirement exists for internal advisory work, but organizational trust, accountability for model risk, and internal governance processes create moderate friction against full AI substitution.
Cost vs. human wageclaude-haiku-4-5-202510012/5Current AI modeling systems require substantial setup, domain customization, and human oversight by skilled statisticians to validate outputs. The total cost (infrastructure, integration, human QA) often approaches or exceeds hiring a statistical consultant for the same advisory work.
Cost vs. human wageclaude-sonnet-53/5AI tools reduce time spent on routine statistical exploration and documentation, but the need for expert oversight and consultation keeps overall cost savings moderate rather than dramatic.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI tools exist for statistical modeling (e.g., AutoML platforms, LLM-based code generation), but deployed systems that autonomously provide advisory-quality statistical guidance to business departments remain rare in production. Most deployed solutions require significant human oversight and cherry-picking of AI suggestions.
Technical feasibility todayclaude-sonnet-52/5AI copilots can suggest statistical methods or generate code, but no deployed product reliably serves as an autonomous advisory resource across departments without human validation and contextual framing.

Recommend investments and investment timing to companies, investment firm staff, or the public.

29

CI 28–31 · exposure 25 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Robo-advisors and AI-assisted portfolio tools have seen steady adoption in wealth management and some institutional settings, but growth has plateaued and human advisors remain dominant in high-net-worth and institutional segments due to regulatory and trust requirements.
Sector adoption velocityclaude-sonnet-53/5Financial services is a fast-adopting sector generally, but for actual investment recommendation and timing (versus back-office analytics), adoption remains cautious with pilots and hybrid human-AI models due to regulatory and liability concerns.
Augmentation potentialclaude-haiku-4-5-202510014/5AI significantly assists human financial advisors by synthesizing market data, running scenario analyses, and suggesting candidate investments; advisors then apply judgment and fiduciary review. This augmentation pattern is common in modern advisory workflows and raises advisor productivity substantially.
Augmentation potentialclaude-sonnet-54/5AI tools significantly enhance financial specialists' ability to analyze data, model scenarios, and surface market signals faster, meaningfully boosting productivity while humans retain final judgment and client relationship responsibilities.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can analyze market data, identify patterns, and generate investment analysis, recommending specific investments and timing requires contextual judgment about risk tolerance, regulatory constraints, and fiduciary duty. Current systems lack the integrated reasoning and liability accountability to perform this end-to-end at 50% time savings with equal quality.
Task automatabilityclaude-sonnet-52/5Generating investment recommendations requires synthesizing market judgment, client-specific risk tolerance, and forward-looking uncertainty that current AI cannot reliably do end-to-end without substantial human oversight; AI can draft analysis but not autonomously deliver defensible investment advice at equal quality.
Adoption barriersclaude-haiku-4-5-202510014/5Investment recommendations are heavily regulated; advisors must be registered and licensed, and fiduciary duties legally rest with humans who must sign off on recommendations. Liability asymmetry (errors harm clients) and regulatory mandates that registered humans certify advice create strong adoption barriers.
Adoption barriersclaude-sonnet-54/5Investment advice is heavily regulated (fiduciary duty, licensing such as Series 65/66, SEC/FINRA oversight), and liability for bad advice creates strong incentives for licensed human sign-off, especially for institutional or complex recommendations.
Cost vs. human wageclaude-haiku-4-5-202510012/5Robo-advisor infrastructure is cheaper than full human advisory at scale, but total cost-per-recommendation including compliance oversight, error remediation, and human sign-off remains comparable to or higher than lower-tier human advisors when liability exposure is factored in.
Cost vs. human wageclaude-sonnet-53/5Automated advisory platforms (robo-advisors) are much cheaper per client than human advisors for basic allocation, but more complex, high-stakes recommendations still require costly human expertise and compliance oversight, keeping overall cost roughly comparable.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI-powered portfolio analytics and recommendation engines exist in production (e.g., robo-advisors), but they operate within narrow, pre-defined rule sets and require human review before recommendations go to end clients. No mainstream system reliably makes independent investment recommendations without human intermediaries in regulated contexts.
Technical feasibility todayclaude-sonnet-52/5Robo-advisors and AI research tools exist and are deployed at scale for basic portfolio allocation, but true investment timing recommendations for companies/public involve nuanced judgment calls that production systems handle narrowly (e.g., simple rebalancing) rather than comprehensively.

Contribute to development of risk management systems.

29

CI 25–32 · exposure 25 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Financial services firms are cautious adopters of AI in risk infrastructure due to regulatory scrutiny, audit requirements, and mission-critical system stakes. Pilots exist but production replacement of specialist roles in system development remains rare; most firms use AI tools as assistants only.
Sector adoption velocityclaude-sonnet-53/5Financial services broadly adopt AI/ML tools at a moderate-to-fast pace, but risk system development specifically remains cautious due to regulatory and model risk concerns.
Augmentation potentialclaude-haiku-4-5-202510014/5AI code generation, documentation, and testing tools materially assist financial risk specialists in development workflows, accelerating routine coding and freeing time for architecture and validation. The human specialist remains in the loop and their productivity is meaningfully elevated by these assistive tools.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist with data analysis, scenario modeling, code generation, and documentation, substantially boosting the productivity of specialists who design these systems.
Task automatabilityclaude-haiku-4-5-202510012/5Risk management system development requires architectural decisions, threat modeling, and integration of complex business logic that demand human judgment and creativity. While AI can assist with code generation and documentation, the core system design and validation processes remain heavily dependent on specialist expertise and cannot yet achieve 50% time savings at equal quality end-to-end.
Task automatabilityclaude-sonnet-52/5System development involves architectural judgment, stakeholder alignment, and integration with organizational risk frameworks that AI cannot autonomously perform end-to-end today.'
Adoption barriersclaude-haiku-4-5-202510014/5Significant regulatory and organizational barriers exist: risk systems are subject to banking, securities, and insurance regulations; internal audit and compliance review is mandatory; and legal liability for system failures falls on the firm and its officers. Automation of system design would require sign-off by licensed professionals, creating hard adoption friction.
Adoption barriersclaude-sonnet-53/5Financial risk systems often require regulatory compliance, model validation, and sign-off from qualified risk professionals, creating moderate institutional and compliance barriers.
Cost vs. human wageclaude-haiku-4-5-202510012/5Risk management system development involves specialized financial expertise commanding high hourly rates ($150–300+). AI tools reduce some coding time but do not eliminate the need for senior domain experts to design, validate, and oversee the system, keeping overall cost per deliverable comparable to or higher than human specialists.
Cost vs. human wageclaude-sonnet-52/5Significant human oversight, domain expertise, and validation are still required, so AI tools reduce but don't dramatically undercut the cost of specialist involvement.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably performs full risk management system development autonomously. AI coding assistants exist and can generate snippets, but production risk systems require domain expertise, regulatory validation, and integration testing that current tools handle only partially and with material gaps.
Technical feasibility todayclaude-sonnet-52/5AI coding assistants and analytics tools help build components of risk systems, but no deployed product independently designs or contributes to full risk management system development reliably.

Devise scenario analyses reflecting possible severe market events.

29

CI 25–32 · exposure 25 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Financial services show moderate pilot adoption of AI for risk analytics, but scenario design remains a high-stakes function with slow, cautious deployment patterns. Most adoption is augmentative (AI assists analysts) rather than autonomous; full automation of scenario devising is rare in production.
Sector adoption velocityclaude-sonnet-53/5Financial services broadly show fast AI adoption, but risk modeling and stress testing functions are more conservative and heavily regulated, leading to moderate, uneven uptake of AI-driven scenario generation.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can significantly augment specialists by rapidly generating candidate scenarios, querying historical databases for precedents, running sensitivity analyses, and flagging parameter inconsistencies, allowing humans to focus on judgment and validation. The human remains accountable but works faster and explores more scenarios.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by rapidly generating base scenarios, historical analogs, and quantitative variations for human risk specialists to refine and interpret, significantly speeding up the analytical process.
Task automatabilityclaude-haiku-4-5-202510012/5Current AI can assist with scenario parameter generation and basic sensitivity analysis, but the task requires domain expertise to conceptualize plausible severe market events, weigh their likelihood, and determine appropriate stress parameters. AI cannot reliably devise novel, contextually sound scenario narratives without substantial human direction and validation.
Task automatabilityclaude-sonnet-52/5Devising credible severe-event scenarios requires judgment about tail risks, correlations, and plausible narratives that current AI cannot reliably generate without extensive human framing and validation, so end-to-end automation with equal quality is not yet achievable.
Adoption barriersclaude-haiku-4-5-202510014/5Regulatory frameworks (Basel III, Dodd-Frank, FCA rules) often require that risk models and stress scenarios be explicitly documented and approved by qualified risk officers; liability for deficient scenarios falls heavily on the institution. The human sign-off requirement and error-cost asymmetry (wrong stress scenarios can mask tail risk) are substantial legal and operational barriers.
Adoption barriersclaude-sonnet-53/5While no license is strictly required to draft scenarios, regulatory expectations (e.g., stress testing under Basel/Dodd-Frank) often require sign-off by qualified risk professionals, creating moderate institutional and compliance friction.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI-assisted tools reduce drafting overhead, but financial risk specialists earn six-figure salaries and must review and validate all output. Integration, domain configuration, and mandatory human oversight make the all-in cost remain comparable to or exceed the human cost for reliable, defensible analyses.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply generate quantitative simulations, but the expert judgment needed to validate and interpret scenarios for regulatory and business relevance still requires costly specialist oversight, keeping all-in costs closer to human levels.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can generate scenario templates and populate parameters via language models and data tools, production systems for enterprise risk scenario design remain limited. No mature off-the-shelf product reliably devises comprehensive scenario analyses meeting regulatory and institutional risk standards autonomously.
Technical feasibility todayclaude-sonnet-52/5Some risk platforms use AI/statistical tools to generate stress scenarios or Monte Carlo simulations, but genuinely novel severe-event narrative construction remains largely human-driven with AI as a minor input, not a reliable deployed capability.

Meet with clients to answer queries on subjects such as risk exposure, market scenarios, or values-at-risk calculations.

26

CI 25–28 · exposure 25 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Financial services firms are piloting AI for internal risk analysis and report generation, but actual client-meeting automation remains rare and limited to low-stakes queries. Adoption velocity in this specific task is slow due to regulatory and reputational risk concerns.
Sector adoption velocityclaude-sonnet-53/5Financial services broadly adopt AI quickly for analytics and drafting, but client-facing advisory meetings remain a slower-adopting, human-centric niche within the sector.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can significantly augment specialist productivity by pre-generating risk analyses, scenario summaries, and values-at-risk calculations that the human specialist reviews and adapts during client meetings. This assistive role improves meeting preparation and response time while keeping the human expert central to client interaction.
Augmentation potentialclaude-sonnet-54/5AI can strongly assist specialists by pre-generating VaR calculations, scenario analyses, and talking points, letting the human focus on interpretation and client rapport during the meeting.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can generate accurate risk analyses and market scenario responses, meeting with clients requires real-time dialogue, nuanced judgment about client risk tolerance, and relationship maintenance that current AI systems cannot reliably perform end-to-end. Client interaction involves trust-building and contextual adaptation that falls short of the 50% time-saving threshold for full automation.
Task automatabilityclaude-sonnet-52/5Client meetings require real-time interpersonal judgment, trust-building, and dynamic Q&A tailored to specific client contexts, which current AI cannot fully replace end-to-end despite being able to draft supporting materials.
Adoption barriersclaude-haiku-4-5-202510014/5Regulatory frameworks in financial services and fiduciary duty standards require that client advice be provided by qualified, licensed professionals who bear accountability. Compliance and client-authorization requirements mean firms cannot fully substitute AI for human risk specialists in client-facing roles.
Adoption barriersclaude-sonnet-54/5Financial risk advice is often subject to regulatory suitability and disclosure requirements, and clients typically expect a licensed, accountable human for sensitive risk discussions, creating meaningful liability and compliance barriers.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI-generated risk explanations are relatively cheap to produce, but integrating them into a client-facing meeting (with oversight, compliance review, and fallback to human specialists) approaches or exceeds the cost of a human meeting. The overhead of ensuring accuracy and liability coverage largely offsets the inference cost savings.
Cost vs. human wageclaude-sonnet-52/5While AI can cheaply generate risk analytics, the human relationship-management and live-meeting component still requires paid specialist time, keeping overall cost comparable to human-only delivery.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably handles complex client meetings on risk topics as a substitute for human specialists. AI chatbots can answer factual questions about risk metrics, but they lack the conversational sophistication, liability accountability, and ability to navigate client-specific concerns that production systems in financial services require.
Technical feasibility todayclaude-sonnet-52/5AI chatbots and copilots exist to answer financial queries, but no deployed product reliably conducts full client risk-advisory meetings with the nuance and relationship management this task demands.

Confer with traders to identify and communicate risks associated with specific trading strategies or positions.

19

CI 11–28 · exposure 13 · augmentation 75 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Financial risk teams use AI-powered risk dashboards and alerting systems, but the conversational, advisory aspects of conferring with traders remain human-driven in most firms. Adoption is slow relative to other financial automation because of regulatory conservatism and the high cost of errors in risk communication.
Sector adoption velocityclaude-sonnet-53/5Financial services is a fast-adopting sector for AI analytics tools, but the interpersonal conferring aspect of this task lags behind pure data-processing automation.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can significantly assist risk specialists by surfacing relevant risk metrics, scenario analysis, and correlation patterns before or during trader conversations, allowing the specialist to focus on dialogue and strategic interpretation. Current generative AI and risk analytics tools already augment this task in practice.
Augmentation potentialclaude-sonnet-54/5AI can pull real-time risk metrics, flag anomalies, and generate pre-meeting risk summaries that significantly enhance the specialist's ability to identify and articulate risks to traders.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can identify and quantify certain structural risks (model outputs, market correlations, VaR calculations), the task fundamentally requires real-time dialogue with traders to understand context, intent, and nuanced strategy variations—something current AI struggles to do autonomously at production quality. The conferential, adaptive aspect of the task cannot be automated end-to-end today.
Task automatabilityclaude-sonnet-51/5This requires real-time, interactive dialogue with traders involving judgment, negotiation, and contextual understanding of live positions, which current AI cannot conduct end-to-end autonomously.
Adoption barriersclaude-haiku-4-5-202510014/5Financial risk management is heavily regulated (Basel III, Dodd-Frank, MiFID II); many jurisdictions require that a qualified human risk officer certify or sign off on risk communications to traders. Liability for miscommunicated risks falls on the firm and individual, creating legal barriers to full automation.
Adoption barriersclaude-sonnet-54/5Risk sign-off and communication with trading desks often carries regulatory and internal governance requirements, and firms retain liability-sensitive human oversight for trading risk decisions.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI inference for risk flagging is cheap, but integrating it into trading workflows and maintaining human oversight of conversational exchanges with traders requires significant cost. A risk specialist's loaded wage for this high-judgment task is likely comparable to or lower than full end-to-end AI deployment costs.
Cost vs. human wageclaude-sonnet-52/5While AI risk analytics tools reduce some computational cost, the human conferring and judgment component still requires a paid specialist, keeping the effective cost ratio close to comparable.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI systems exist to flag and analyze portfolio risks and model alerts, but they do not reliably conduct multi-turn dialogues with traders to identify strategy-specific risks and negotiate communication in production settings. Deployed products perform narrow risk alerting, not the collaborative consultation this task requires.
Technical feasibility todayclaude-sonnet-51/5No deployed product independently confers with traders to identify and communicate risk in a conversational, relationship-based advisory capacity; this remains a human-led interaction.

Related occupations — Business & Financial Operations

How to read this

A high substitution score does not mean this job disappears — it means a large share of its current tasks face replacement pressure, so the mix of tasks is likely to change. High augmentation alongside substitution typically means the occupation reorganizes around the protected tasks. Wide confidence intervals mean the rater panel disagreed: treat those scores as open questions, not verdicts.

What would change this score

New model capabilities (automatability, feasibility), falling inference costs (cost ratio), regulation and licensing shifts (barriers), and measured sector adoption (velocity) all re-enter at every index release. Each release is recomputed, versioned and kept queryable — scores are claims with a date on them, not permanent labels.