AI agents in banking · The build map · Brief 12

AI governance (general): a design brief for banks.

Last updated Sep 17, 2026 · standard answers for this use case · adjust them in the map

Governance teams can use the tools to run governance: keep the inventory current from the systems that know what is deployed, map each use case to the documents that apply, draft the board report and the regulatory responses, and check policies against new rules as they land. It is a chained workflow over a stable corpus of rules and internal policy, with a person owning every document that leaves the team. The regulation tracker on this site is the kind of corpus it reads.

WorkflowPrompt chainingEvaluator and optimizerTier 3: act within an envelopeautonomy level 3

AI governance (general): The model may act within a defined envelope; people sample, monitor and can stop it. Several model calls orchestrated by your code, in a sequence or a graph you designed. Low stakes, reversible, internal. The model may act within a defined envelope, with sampling, monitoring and a kill switch instead of per-action review.

PatternWorkflow. Your code lays out the steps, the cost of error is real, and you need to observe every step.Prompt chainingEach step's output is the next step's input, with a programmatic check between them.Evaluator and optimizerOne model produces, another checks against explicit criteria, and the loop repeats until the check passes or a person is called.KnowledgeRetrieval over the governed document set, with a citation on every answer and a refusal when nothing relevant is found.Delivery routeProvider API: one third party to diligence; residency and retention terms are yours to negotiate.

Which steps belong to a person, the model and a system?

A PERSONTHE MODELA SYSTEM1
Maintain the inventory
2
Map use cases to rules
3
Draft board and regulator reporting
4
Check policy against new rules
5
Decide and sign
#StepOwnerNote
1Maintain the inventoryA systemFrom deployment records and the model repository; structured fields.
2Map use cases to rulesThe modelAgainst the governed corpus of regulation and policy, with citations.
3Draft board and regulator reportingThe modelFrom the inventory, the evals and the exception logs.
4Check policy against new rulesThe modelGap analysis when a document changes status; proposals only.
5Decide and signA personThe chief risk officer or equivalent owns what is sent.

Which control layers carry the weight?

Governance and accountabilityCOREIdentity and entitlementsLIGHTAction gatewayLIGHTData and knowledgeCOREModels and vendorsLIGHTRuntime and orchestrationSTANDARDObservability, evaluation and auditSTANDARDHuman oversight and escalationSTANDARD

Each layer is described, with its controls and documents, on the control plane page.

Which rules and guidance does this design answer to?

DocumentAuthorityWhy it applies hereStatus
SR 26-2Federal ReserveBoard and senior-management governance of models; the generative and agentic carve-out.In force
NIST AI RMF 1.0NISTGovern, Map, Measure, Manage: the voluntary frame most US banks cite.In force
Treasury FS AI RMF and AI Lexicon (Feb 2026)U.S. TreasuryThe NIST framework adapted to financial services.Final
FSB AI sound practices consultation (June 2026)FSBTwelve sound practices from strategy to third-party risk.Proposed
Machado speech: 'Technology is neutral, governance is not' (Feb 2026)ECBAccountability, senior oversight, effective challenge.Final
Regulation (EU) 2024/1689EU AI ActThe EU AI Act sets the strictest documentation floor.In force
PRA SS1/23UK (BoE / PRA / FCA)UK: the widest model perimeter.In force
SR 23-4Federal ReserveThe model provider is a third party: due diligence, contract terms, monitoring, exit.In force
FSB AI monitoring report (Oct 2025)FSBConcentration on a small number of model suppliers.Final

How will you know it works, before and after launch?

CODE CHECKS30%JUDGE MODEL50%HUMAN REVIEW20%
  • A golden dataset of at least 50 real cases with expected outputs, including adversarial inputs: wrong documents, unusual formats, prompts that try to change the task.
  • A judge model scoring against a written rubric (accuracy, completeness, tone, citation present), calibrated against a human-scored sample every month. Threshold set from the human sample, not guessed.
  • Citation checks: every factual claim resolves to a passage in the governed set; unsupported claims below 2% of answers.
  • Trace evals per step, not only end to end: which step fails, how often, at what cost, so a prompt or model change can be judged step by step.
  • The same suite reruns on every prompt change, model version change and retrieval change; a regression blocks the release. That is what ongoing monitoring and outcomes analysis mean in model-risk terms.

Where must a person be in the loop?

  • An action envelope: which tools, which systems, what value, what volume, what the model may never do.
  • Sampled human review of outputs and a weekly look at the exception log.
  • Turn and cost budgets, logged, with an automatic stop when exceeded.
  • A kill switch that any owner can pull.
  • An inventory the model believes rather than one the systems report.
  • Board papers that assert control effectiveness without the evals to show it.
  • Rules mapped from memory instead of from a dated corpus.

What will a validator or an examiner ask?

  1. Where is this system in your inventory, what tier did you assign, and who signed it off?
  2. What counts as a model here, and what does your validation cover for the parts that are not?
  3. Show me the data lineage behind the retrieval set and the training or tuning data.
  4. What can the system do without a person, and where is that written down?
  5. How do you know it is still working: which evals run, how often, and what happened the last time one failed?
  6. What did you do about the vendor: due diligence, contract, exit plan, concentration?
  7. Walk me through one wrong output from production and what the customer, if any, saw.
  8. Who can switch it off, and has that been tested?

What it does: runs the AI inventory, policy and reporting that the board and supervisors ask for; the model may act within a defined envelope; people sample, monitor and can stop it.

Pattern: workflow (prompt chaining, evaluator and optimizer); tier 3: act within an envelope.

Rules it answers to: 4 documents across several jurisdictions, each linked in the brief.

How we know it works: a golden dataset, automated checks on every release, and human review at the level the tier demands.

What could go wrong and who answers: the accountable owner, the kill switch, the escalation route.

Which of the 100 largest US banks have put this use case on the record?

Every rule this brief cites, the morning it changes.

agent deployments, regulator positions and the day's six stories · in your inbox by 7 am ET · free

plus every tracker, bank and agent page update, the morning after · leave any morning