AI use cases in finance

Most AI use cases break down the same way. There is the data it learns from or works on, and whether that data is right for the job. There is the model or method that fits the task, which is often simpler than the hype suggests. There is the task itself, what it actually does, and whether it supports a person or acts on its own. And there is evaluation and testing, how you know it works and where the weak point is. Run any use case through those four lenses and it stops being a vague "AI project" and becomes something you can size, test and govern.

You will notice this catalogue does not sort use cases into risk tiers. That is deliberate. How material a use case is depends on how you deploy it: who it affects and how many, whether a person genuinely re-decides or only approves, whether the action can be reversed, and your own regulatory and fairness exposure. The same use case can be a high-materiality use in one firm and a lower one in another. So each page gives you the four lenses and the questions to ask; you judge the materiality against your own deployment, and let the controls follow. For how that rating is done, see use-case risk and materiality.

This is the finance view. For other industries, see AI use cases by industry, or try the interactive AI use-case explorer. The finance catalogue is below, by area. Each use case links to its own page as it is written. Last reviewed 24 October 2026; to add or correct one, write to gary@quaintitative.com.

Banking

• Credit decisioning. Score default risk and decide or price a loan.
• Transaction fraud detection. Flag, and sometimes block, suspicious payments in near real time.
• AML alert triage and sanctions screening. Surface and summarise suspicious activity for investigators.
• KYC and onboarding due diligence. Verify identity and assess customer risk at onboarding.
• Customer churn and next-best-action. Predict who is likely to leave and what to offer them.
• Branch and call-centre service assistant. Answer staff and customer questions from policy and product documents.
• Collections prioritisation. Rank overdue accounts for action.

Insurance

• Claims triage and straight-through processing. Read a claim, assess it, and pay, request more, or escalate.
• Underwriting risk assessment and pricing. Assess and price risk from application and third-party data.
• Claims fraud detection. Flag suspicious claims for investigation.
• Damage assessment from photos. Estimate damage and cost from images.
• Policy and claims letter drafting. Draft customer correspondence from templates and case data.
• Document intake and extraction. Pull structured data from forms and documents.
• Catastrophe and loss forecasting. Model expected losses from exposure and event data.

Capital markets and investment

• Price, demand and volatility forecasting. Predict market variables to inform decisions.
• Trade and communications surveillance. Detect market abuse and misconduct in trading and messages.
• Algorithmic and execution strategy. Decide how and when to execute orders.
• Position and exposure monitoring. Flag limit breaches and unusual exposure.
• Research and earnings-call summarisation. Summarise filings, calls and news for analysts.
• ESG and news signal extraction. Extract sentiment and signals from text.
• Portfolio construction and rebalancing. Propose allocations and rebalancing.

Fintech

• Alternative-data credit scoring. Score thin-file borrowers from non-traditional data.
• Real-time payment fraud scoring. Score each payment for fraud as it happens.
• Onboarding identity verification. Match a document to a selfie and verify identity.
• Personal-finance insights assistant. Answer customer questions about their own money.
• Spend categorisation. Label transactions by category.
• Chargeback and dispute automation. Read, classify and summarise disputes.
• Robo-advisory recommendations. Recommend portfolios and actions to retail investors.

Cross-cutting (any finance firm)

• Internal knowledge and policy assistant. Answer staff questions from internal documents, with citations.
• Document processing. Extract and classify data from invoices, forms and contracts.
• Meeting transcription and notes. Transcribe and summarise meetings.
• Email drafting and inbox triage. Draft replies and route incoming mail.
• Code assistant. Help developers write and review code.
• Talk-to-your-data analysis. Answer analytical questions over a dataset in plain language.
• Enterprise search. Find documents and answers across internal systems.
• Resume screening. Extract, classify and rank job candidates.
• Report and spreadsheet generation. Draft reports and summaries from data.
• IT and security anomaly detection. Flag unusual system or security events.