Financial Services
Deploying Claude for Financial Services: Banks, Asset Managers, Insurers
What Claude can do for banks, asset managers and insurers, how regulated firms deploy it, and what SR 26-2, the NAIC Model Bulletin and FINRA say, from a Claude Partner Network, Select partner.
Claude for Financial Services is Anthropic's offering for banks, insurers, asset and wealth managers and fintech firms: Claude's models and apps, financial data connectors and implementation support. Firms run it through Claude's own apps and API, or inside AWS, Google Cloud or Microsoft Azure. Claude assembles the work; a named professional keeps every judgment call.
what claude does in financial services
Claude for finance teams: what it does, and where a person stays accountable
Anthropic's financial services page states the pattern plainly: Claude does the assembly, and the professional keeps control of the judgment calls. Make that a written rule: for each use case, record what Claude produces, who reviews it, and which decision that person signs. If no one can be named, the use case is not ready.
The middle column paraphrases Anthropic's page; the right-hand column is our recommendation, not a regulatory statement.
Sources, connectors and checking the work
Anthropic says Claude can link figures and claims back to the underlying data and documents, and lists connectors to LSEG, FactSet, S&P Global, Morningstar, PitchBook, Moody's, Databricks, Snowflake and others, plus your internal data. A citation helps the reviewer; it does not replace the review.
For KYC and AML work, see our governed AI for KYC/AML.
| Use case | What Claude does (per Anthropic's financial services page) | Who stays accountable |
|---|---|---|
| Investment banking pitch materials | Builds pitch books, comps tables and CIMs in PowerPoint and Excel. | The banker owns the narrative and the valuation calls. |
| Commercial banking credit memos | Pulls spreads, runs ratios and structures credit memos in Excel. | The credit officer owns the credit decision. |
| Asset management portfolio reporting | Drafts IC memos, performance decks and portfolio reviews from holdings data. | The portfolio manager owns the investment thesis. |
| Insurance actuarial review | Reviews actuarial workbooks and filings, flags anomalies, drafts filing narratives. | The actuary retains sign-off on reserves. |
| KYC screening and AML case preparation | Anthropic lists KYC screening agent templates and agents that prepare KYC work for a team's review. | A named analyst reviews each case; any SAR filing decision stays with your BSA/AML function. |
| Engineering and legacy code | Claude Code builds, fixes and updates software, including legacy systems. | The engineer who merges the change owns it, through normal code review. |
context
Claude for financial advisors: what Anthropic offers and who stays accountable
Claude for financial advisors means using Claude to prepare work that a licensed professional then reviews and owns. Anthropic offers Claude for Financial Advisors, which its page says connects to custodian, portfolio, planning and CRM systems, with a plugin covering onboarding, meeting prep, portfolio reviews, proposals and compliance reviews, with review steps and an audit trail built in. The same page says, of its finance use cases in general, that Claude does the assembly and the professional keeps control of the judgment calls. For an advisory firm that points to preparation and drafting, with the advisor deciding what reaches a client. Kriv AI does not give investment advice. We recommend that every client recommendation stays with a named, licensed person. We recommend that your compliance function approves each use case and records who reviews the output.
deployment options
Deployment options for regulated firms: Claude apps, the API, AWS, Google Cloud and Azure
Anthropic lists three ways to deploy Claude for Financial Services: ready-made apps such as Claude Enterprise, Claude Code for engineering teams, and the API for building Claude into your own risk, KYC and underwriting systems. It also says Claude runs on AWS, Google Cloud and Azure. For a regulated firm the question is who operates inference and under whose terms your prompts are processed.
This page quotes no Claude plan or usage prices; see Anthropic's pricing page and each cloud provider's pricing page. See also our guide to adding a Claude skill.
How to choose between them
Start from where your vendor-risk approvals and cloud agreements already sit: a firm that has approved AWS may find Bedrock the shorter review; a firm standardized on Azure may look at Foundry first. Then confirm in the provider's current documentation who operates inference, which data-use terms apply, and whether the model you need is available in a region your policy allows. Record the answers in the use case's inventory entry, and re-read the pages in our sources list before you sign, because provider documentation changes.
| Option | Who operates it (per the provider's documentation) | What the documentation says about your data |
|---|---|---|
| Claude apps (for example Claude Enterprise) and the Claude API, direct from Anthropic | Anthropic. | Anthropic's privacy center: by default it will not use inputs or outputs from its commercial products to train its models. Its financial services page: 'On Team and Enterprise plans, we don't train our models on your data by default.' |
| Claude in Amazon Bedrock | AWS. Anthropic's documentation: 'zero operator access (Anthropic personnel have no access to the inference infrastructure)'. | Access runs through your AWS account and IAM roles; Bedrock sets access criteria for each Claude model. |
| Claude Platform on AWS | Anthropic, with AWS providing authentication, IAM access control and AWS Marketplace billing. | 'Anthropic is the data processor for inference inputs and outputs'; an inference_geo parameter can pin inference to a geography. |
| Claude on Google Cloud (Gemini Enterprise Agent Platform, Model Garden) | Google Cloud, as 'fully managed and serverless models as APIs'; Google notes Anthropic models are not a Google product and carry separate terms. | Optional request-response logging for 30 days. Anthropic notes model availability varies by region. |
| Claude in Microsoft Foundry | Anthropic, on Azure or Anthropic infrastructure depending on the hosting option chosen; billed through the Azure Marketplace. | 'For deployments hosted on Azure, prompts and completions remain within Azure.' A US Data Zone Standard deployment keeps inference within the United States. |
governance and model risk
Governance and model risk: what SR 26-2, the NAIC bulletin and FINRA say
What each document says about AI, quoted from the primary source.
SR 26-2: banks, and why generative AI sits outside its scope
On April 17, 2026 the Federal Reserve issued SR 26-2, transmitting revised model risk management guidance from the Federal Reserve, the OCC and the FDIC that 'supersedes and replaces' SR 11-7 and SR 21-8, the 2021 interagency statement on model risk management for BSA/AML systems.
The key line for Claude is footnote 3: 'Generative AI and agentic AI models are novel and rapidly evolving. As such, they are not within the scope of this guidance.' It adds that 'a banking organization's risk management and governance practices should guide the determination of appropriate governance and controls' for tools it does not cover, and the guidance says it 'does not set forth enforceable standards or prescriptive requirements.' So no Claude rollout is 'SR 26-2-compliant'. A bank can still write Claude into its own practices on purpose: an inventory entry, a named owner, a risk tier per use case, and monitoring. See AI model risk management under SR 26-2.
The NAIC Model Bulletin: insurers
The NAIC Model Bulletin on the Use of Artificial Intelligence Systems by Insurers was adopted by the NAIC Executive Committee and Plenary on December 4, 2023. It is a model: each state's insurance department issues its own version, and its drafting note says states may have adopted different definitions.
The model text defines Generative AI as a class of AI Systems. It says insurers 'are expected to develop, implement, and maintain a written program (an "AIS Program") for the responsible use of AI Systems' that make or support decisions related to regulated insurance practices, and covers due diligence on third-party AI systems. See NAIC AI model governance for insurers and AI for insurers and payers.
FINRA Regulatory Notice 24-09: broker-dealers
FINRA Regulatory Notice 24-09 (June 27, 2024) says FINRA's rules are 'intended to be technology neutral' and continue to apply when member firms use generative AI, and that the Notice 'does not create new legal or regulatory requirements'. Its example: where generative AI is part of a supervisory system, policies 'should address technology governance, including model risk management, data privacy and integrity, reliability and accuracy of the AI model.'
How we apply model-risk discipline to a Claude rollout
Any Claude deployment touching a regulated decision should get the same model inventory and validation treatment our model risk management practice applies more broadly, with every interaction logged for audit.
We do not certify alignment, and whether you comply with a law is for your counsel. We write the rules to the policy and frameworks you already answer to, such as SR 26-2, your state's version of the NAIC Model Bulletin, FINRA's notices and the NIST AI RMF. See AI governance consulting for financial services.
where kriv fits
Where Kriv AI fits in a Claude for financial services rollout
We bring the governance approach behind our work in regulated industries, including model risk management, to financial services. We don't have a named financial-services case study to point to yet; what we do have is our status as a Claude Partner Network, Select partner and a security and compliance program built for regulated industries. Engagements are scoped and priced after a short discovery call.
Kriv AI has been named a Select partner in the Claude Partner Network. Our status is 'Claude Partner Network, Select partner · Services Track'. Program details at claude.com/partners; written confirmation from the Claude Partner Network team available on request. We are not Anthropic and do not speak for it.
Kriv AI sends a senior practitioner who both scopes and delivers: no handoffs, no bench, no discovery theater. That person helps you pick first use cases with a named owner each, match a deployment path to your cloud and vendor approvals, write inventory entries, risk tiers and sign-off points into your policy, and train your people. Use-case registers and framework mappings can be prepared on request. See also our guide to a first Claude Code session.
Training builds on Claude Enterprise rollout and training and Claude Code training for engineering teams.
Related credentials: Microsoft 'Solutions Partner for Cloud & AI Platforms · Partner ID 7036589', 'AWS Marketplace seller, 59 live listings', and Databricks partner.
For banks, asset and wealth managers, insurers and fintech firms whose compliance or model-risk function approves each Claude use case. Not for a team that only wants seats switched on.
See how engagements are scoped and priced, or read about Kriv AI in the Claude Partner Network.
Sources
Cited sources
- Anthropic, 'Financial services | Claude by Anthropic' (fetched 2026-10-20 UTC, batch13/sources/claude-fs.html): use cases by role, 'Claude does the assembly, and the professional keeps control of the judgment calls', three deployment paths, 'Claude runs on AWS, Google Cloud, and Azure', connector list, agent templates for KYC screening, 'On Team and Enterprise plans, we don't train our models on your data by default'
- Anthropic, 'Financial services' (claude.com/solutions/financial-services, reached from anthropic.com/solutions/financial-services), read 2026-11-10
- Anthropic Privacy Center, 'Is my data used for model training?' (fetched 2026-10-20 UTC, batch13/sources/privacy-train.html)
- Anthropic, 'Plans & pricing | Claude by Anthropic' (fetched 2026-10-20 UTC, HTTP 200; linked, not quoted)
- Claude Platform Docs, 'Claude in Amazon Bedrock' (fetched 2026-10-20 UTC, batch13/sources/doc-claude-in-amazon-bedrock.html)
- Claude Platform Docs, 'Claude Platform on AWS' (fetched 2026-10-20 UTC, batch13/sources/doc-claude-platform-on-aws.html)
- Claude Platform Docs, 'Claude in Microsoft Foundry' (fetched 2026-10-20 UTC, batch13/sources/doc-claude-in-microsoft-foundry.html)
- Claude Platform Docs, Claude on Google Cloud's Agent Platform (URL slug still 'claude-on-vertex-ai'; fetched 2026-10-20 UTC, batch13/sources/doc-claude-on-vertex-ai.html)
- Google Cloud Documentation, 'Anthropic's Claude on Google Cloud models | Gemini Enterprise Agent Platform' (requested at cloud.google.com/vertex-ai/generative-ai/docs/partner-models/claude, which redirected here; fetched 2026-10-20 UTC, batch13/sources/vertex-claude.html)
- Federal Reserve, SR 26-2 'Revised Guidance on Model Risk Management', April 17, 2026 (letter; fetched 2026-10-20 UTC, batch13/sources/sr2602.htm): 'supersedes and replaces SR letter 11-7 ... and SR letter 21-8'
- SR 26-2 Attachment, 'Supervisory Guidance on Model Risk Management' (PDF; fetched 2026-10-20 UTC, batch13/sources/sr2602a1.pdf): section I ('does not set forth enforceable standards or prescriptive requirements'), footnote 3 (generative and agentic AI not within scope)
- NAIC Model Bulletin: Use of Artificial Intelligence Systems by Insurers, adopted December 4, 2023 (PDF; fetched 2026-10-20 UTC, batch13/sources/naic.pdf): Section 2 definitions (Generative AI), Section 3 (AIS Program), Section 4 item 4.0 (Third-Party AI Systems and Data)
- FINRA Regulatory Notice 24-09, published June 27, 2024 (fetched 2026-10-20 UTC, batch13/sources/finra-24-09.html)
- Claude Partner Network program page (program details only; not a status listing, per cmo ruling)
Related resources
Continue exploring
Straight answers
Frequently asked questions about Deploying Claude for Financial Services: Banks, Asset Managers, Insurers
What is Claude for Financial Services?
Claude for Financial Services is Anthropic's offering for banks, insurers, asset and wealth managers and fintech firms: Claude's models, apps, financial data connectors and implementation support. Kriv AI is not Anthropic. As a Claude Partner Network, Select partner we help regulated firms deploy and govern it: use cases, named owners, a deployment path, and controls written into your own policy.
What can Claude for finance teams do, and what should it not decide?
Anthropic lists pitch books and comps for bankers, credit memos and spreads for commercial lenders, IC memos and portfolio reviews for asset managers, and actuarial and filing review for insurers. It says Claude does the assembly and the professional keeps the judgment calls. Put that in writing: the credit decision, the valuation, reserve sign-off and any SAR filing belong to a named person.
Is Claude AI secure enough for financial data?
It depends on where Claude runs and the terms you sign. Anthropic says it does not train on inputs or outputs from its commercial products by default. Its documentation says Claude in Amazon Bedrock runs with zero Anthropic operator access, and that on Azure-hosted Foundry deployments prompts and completions remain within Azure. Your own security review decides.
Is Claude AI compliant for financial services use?
No foundation model is 'compliant' on its own. Compliance comes from how the deployment is governed: the bank's own model risk discipline (SR 26-2 places generative AI outside its scope), audit logging and a defined data-handling policy.
How does the NAIC Model Bulletin apply to insurers using Claude?
Where a state has issued it, the bulletin covers AI Systems that make or support decisions related to regulated insurance practices, and its definitions include Generative AI. The model text says insurers are expected to maintain a written AIS Program and addresses due diligence on third-party AI systems. Read your own state's version, since definitions can differ by state.
Can we run Claude inside our existing AWS, Google Cloud or Azure account?
Yes, according to Anthropic's documentation, which covers Claude in Amazon Bedrock, Claude Platform on AWS, Claude on Google Cloud's Agent Platform and Claude in Microsoft Foundry. Each has its own operator, billing and data terms: Bedrock is operated by AWS, while Claude Platform on AWS is operated by Anthropic and billed through AWS Marketplace. Check model availability in your region first.
What does Claude AI implementation cost for a bank or asset manager?
Every engagement is scoped and priced after a short discovery call. Our fee covers our work only. Claude plan prices are Anthropic's and are on its pricing page; if you run Claude through Amazon Bedrock, Google Cloud or Microsoft Foundry, that provider bills usage under your agreement with it. We quote neither here, because both change.
Does Kriv AI have a named financial services client?
Not yet, and we're not going to pretend otherwise. We don't have a named financial-services case study to point to yet, and we will not name any client without written consent. What we bring is a senior practitioner who both scopes and delivers, Claude Partner Network, Select partner status, and governance written to the frameworks you already answer to.
Talk to the team that would do the work
Bring your requirements to a working session with the person who'll actually deliver.
Book a Discovery Call