What should a bank look for in enterprise AI assistants software?
Enterprise AI assistants are the general-purpose large-language-model products banks license for their staff and developers: chat assistants in the office suite, coding assistants, and cloud platforms that give access to several models through an API. They are the one category where list prices are usually published, per user per month or per token. The questions that separate them are data-use terms, where data is processed and stored, admin and audit controls, integration with the bank's identity and document systems, and the total cost at scale.
Observations from how these products are sold and regulated, not advice.
01
Data-use and retention terms
Check that the enterprise terms exclude your prompts and files from model training, how long they are retained, and where they are processed.
02
Identity, permissions and audit
An assistant that searches internal documents must respect existing permissions. Check single sign-on, role controls and exportable audit logs.
03
Seat price versus usage price
Per-seat assistants have predictable bills; per-token platforms scale with use. Published list prices are a ceiling for large banks, which negotiate.
04
Coding assistants and secure development
Code-generation tools raise secure-development and licence questions; NIST's SSDF profile for generative AI and NY DFS's 2026 letter address them.
Which rules apply to a bank using enterprise AI assistants software?
| Rule | Why it applies | Documents |
|---|---|---|
| Third-party and concentration risk | Licensing a frontier model or cloud AI platform is a third-party relationship; supervisors have flagged concentration in a few providers. | OCC Bulletin 2023-17 SR 23-4 |
| Generative AI risk management | SR 26-2 leaves generative and agentic AI outside its scope and points banks to broader governance; NIST's Generative AI Profile lists the risks to manage. | SR 26-2 NIST AI 600-1 (Generative AI Profile) |
| AI-generated code | NY DFS's May 2026 letter and NIST SP 800-218A address controls over code written with AI assistants. | DFS Frontier AI Models Industry Letter (May 2026) NIST SP 800-218A (SSDF profile for generative AI) |
Documents link to the Regulation tracker. By use case: Generative & agentic AI · AI-generated code & coding agents · Third-party & vendor AI.
How do the enterprise AI assistants vendors compare?
| Vendor | What it does | Deployment | Pricing model | Published price | US bank customers* | Founded · ownership |
|---|---|---|---|---|---|---|
| Microsoft 365 Copilot | AI assistant inside Word, Excel, Outlook and Teams, grounded in a firm's Microsoft 365 data, plus Copilot Studio for building agents. | SaaS add-on to a qualifying Microsoft 365 plan (Microsoft commercial cloud and select sovereign clouds) | Per-user monthly licence added to a qualifying Microsoft 365 plan, paid yearly or monthly; agents in Copilot Studio billed through prepaid Copilot Credit packs or pay-as-you-go | $30.00 per user per month, paid yearly (annual subscription); a qualifying Microsoft 365 plan is required separately (Microsoft 365 Copilot (enterprise)) | 25 | 2023 · Product of Microsoft |
| ChatGPT Enterprise (OpenAI) | OpenAI's workplace ChatGPT with admin controls and no training on business data by default, plus API access to GPT models. | SaaS workspace (ChatGPT Business and Enterprise) and API (cloud); OpenAI models are also sold through Amazon Bedrock and power Microsoft 365 Copilot | Per-seat monthly subscriptions for ChatGPT Business (standard and premium seats, billed annually or monthly); ChatGPT Enterprise on annual, custom-priced contracts with volume discounts; API billed per token | $20 per user per month, billed annually ($25 per month if billed monthly) (ChatGPT Business — Standard seat) | 17 | 2015 · Product of OpenAI Group PBC, a public benefit corporation controlled by the nonprofit OpenAI Foundation |
| Claude (Anthropic) | Anthropic's Claude models and workplace plans (Team, Enterprise), sold direct and through AWS and Microsoft. | SaaS workspace (Claude Team and Enterprise) and API (cloud); Claude models are also sold through Amazon Bedrock and run inside Microsoft 365 Copilot | Per-seat subscriptions for Claude Team; Claude Enterprise at a per-seat fee plus usage billed at API rates; API billed per million input and output tokens by model | US$20 per seat per month, billed annually, 20-seat minimum, plus usage at API rates (Claude Enterprise) | 24 | 2021 · Product of Anthropic PBC, a private public benefit corporation |
| Gemini Enterprise (Google Cloud) | Google Cloud's workplace AI app and agent platform on Gemini models, connecting to Google Workspace, Microsoft 365 and business data. | SaaS on Google Cloud (Gemini Enterprise app, per-seat or pay-as-you-go), with agents built on Gemini Enterprise Agent Platform | Per-seat monthly subscription by edition (Business, Standard, Plus), or pay-as-you-go usage pricing with no seat fee for organisations with 20 or more seats | From $21 USD per seat per month; up to 300 seats (Gemini Enterprise app — Business edition) | 11 | 2025 · Product of Google Cloud, part of Alphabet |
| Amazon Bedrock (AWS) | AWS service giving one API to foundation models from Amazon, Anthropic, OpenAI, Meta and others, plus AgentCore for running agents. | API (cloud) on AWS, with in-Region, geographic and global cross-Region inference options | Pay-as-you-go per million input and output tokens, set per model and varying by region, cross-Region routing and service tier; batch and reserved options; AgentCore billed on consumption | $0.30 input / $2.50 output per 1M tokens; Standard tier, global cross-Region inference (Amazon Bedrock — Amazon Nova 2 Lite) | 11 | 2023 · Product of Amazon Web Services, part of Amazon |
| Microsoft Foundry (formerly Azure AI Foundry) | Azure platform for building AI apps and agents on models from OpenAI, Anthropic, Meta and others; includes Azure OpenAI. | API and platform service (cloud) on Azure, with Global, Data Zone (EU or US) and Regional deployment types | Consumption-based: each Foundry service has its own billing model; models are billed per million input and output tokens by model and deployment type (Global, Data Zone, Regional), with batch and hourly provisioned-throughput options | $2 input / $10 output per 1M tokens (cached input $0.20); Data Zone deployment $2.20 / $11 (Azure OpenAI in Foundry Models — GPT-6 Sol (short context), Global deployment) | 3 | 2024 · Product of Microsoft |
| GitHub Copilot | AI coding assistant and coding agent in the IDE, CLI and GitHub.com, with Business and Enterprise plans billed per seat plus AI credits. | SaaS (GitHub cloud) with IDE, command-line and GitHub.com integrations | Per granted seat per month, with a monthly allowance of GitHub AI Credits pooled across the organisation (Copilot Business and Enterprise); additional usage billed at $0.01 per credit under admin-set budgets. Since October 28, 2026, GitHub Enterprise Cloud licensees can instead get Copilot access included at a $0 seat charge, with all usage metered from the first request against the enterprise's AI budget and no pooled credit allowance; the same access costs $10 per user per month (Copilot Standalone) for users without a GitHub Enterprise licence. | $19 USD per granted seat per month; includes 1,900 GitHub AI Credits per user per month (GitHub Copilot — Copilot Business) | 11 | 2022 · Product of GitHub, which Microsoft |
* Banks among the 120 largest US banks on this site publicly named as customers, each sourced on the vendor's page. Prices are shown only where the vendor publishes them for the AI product itself; a price for another product (a credit score, a notary plan, contact-centre seats) is listed on the vendor's page, not here. More vendors in this category are being researched.
NASDAQ: MSFT · founded 2023
Microsoft 365 Copilot
Sells banks a per-user Copilot licence layered on the Microsoft 365 contracts they already hold, with Copilot Studio agents billed by usage; the pitch is reach into existing email, documents and Teams data under Microsoft's enterprise data terms.
Banks among the 120: BofA, Wells Fargo, BNY, M&T, Northern Trust, BMO, UBS, SouthState, First Citizens, United Community, Bank of Hawaii, First Merchants, TowneBank, Popular, OceanFirst, Western Alliance, Fifth Third, Citizens, State Street, Live Oak, Santander, PNC, KeyBank, HSBC, Barclays US
Private · founded 2015
ChatGPT Enterprise (OpenAI)
Sells banks ChatGPT seats for employees and API access to its models, and in 2026 added on-site engineering deals, a financial-services edition built with investment-bank design partners, and a separately funded deployment company.
Banks among the 120: Morgan Stanley, BNY, Customers Bank, SouthState, JPMorgan, Goldman, First Interstate, Synchrony, U.S. Bank, TD, UBS, Santander, Citizens, OceanFirst, HSBC, Bank OZK, Deutsche Bank
Private · founded 2021
Claude (Anthropic)
Sells banks Claude seats and API access directly and through cloud marketplaces, and in 2026 added embedded engineering with Goldman Sachs and FIS, restricted cyber-defence access to its Mythos model for a few large banks, and a services firm co-founded with Goldman Sachs.
Banks among the 120: Goldman, JPMorgan, BankUnited, Flagstar, SoFi, Deutsche Bank, OceanFirst, BofA, Citi, Morgan Stanley, Schwab, Barclays US, Wells Fargo, BNY, Citizens, Fifth Third, BMO, Santander, Live Oak, Optum Bank, U.S. Bank, Axos Bank, Raymond James, Amex
NASDAQ: GOOGL · founded 2025
Gemini Enterprise (Google Cloud)
Sells banks a per-seat AI app for employees plus an agent-building platform on Google Cloud, and since August 2026 a financial-services edition for capital markets and corporate banking built with bank design partners.
Banks among the 120: Wells Fargo, BNY, Citi, Zions, Goldman, USAA, Santander, U.S. Bank, Optum Bank, Huntington, Deutsche Bank
NASDAQ: AMZN · founded 2023
Amazon Bedrock (AWS)
Sells banks pay-per-use access to many foundation models inside the AWS accounts and regions where their data already sits, plus agent infrastructure; banks often reach it through software vendors, such as Fiserv and Nasdaq Verafin, that build on it.
Banks among the 120: BankUnited, FNBO, First Interstate, USAA, Synchrony, Citizens, U.S. Bank, SouthState, Bank OZK, Flagstar, SoFi
NASDAQ: MSFT · founded 2024
Microsoft Foundry (formerly Azure AI Foundry)
Sells banks consumption-priced access to many foundation models, agent hosting and governance tooling inside the bank's own Azure subscription, under Microsoft's data protection terms; banks also reach it through software vendors such as Fiserv that build on it.
Banks among the 120: Northern Trust, BNY, Ally
NASDAQ: MSFT · founded 2022
GitHub Copilot
Sells banks per-seat Copilot Business and Enterprise licences for developers, each including a monthly allowance of usage-based AI credits since June 2026, with coding agents from GitHub, Anthropic and OpenAI running inside GitHub.
Banks among the 120: Goldman, TD, Citizens, Regions, Northern Trust, Fifth Third, CIBC, Amex, Deutsche Bank, U.S. Bank, BNY
How much does Microsoft 365 Copilot or ChatGPT Enterprise cost a bank?
List prices for enterprise assistants are published per user per month on the vendors' own pricing pages, and cloud model platforms publish per-token prices; each vendor page here quotes the list price with the date it was checked. Large banks negotiate enterprise agreements below list.
Do banks let employees use ChatGPT?
Many large banks license enterprise versions of these assistants under contract terms that exclude bank data from model training, and several have built internal assistants on top of the same models. The bank pages on this site record which banks have said so.
Which enterprise AI assistants vendors are compared here?
Microsoft 365 Copilot (AI assistant inside Word, Excel, Outlook and Teams, grounded in a firm's Microsoft 365 data, plus Copilot Studio for building agents); ChatGPT Enterprise (OpenAI) (OpenAI's workplace ChatGPT with admin controls and no training on business data by default, plus API access to GPT models); Claude (Anthropic) (Anthropic's Claude models and workplace plans (Team, Enterprise), sold direct and through AWS and Microsoft); Gemini Enterprise (Google Cloud) (Google Cloud's workplace AI app and agent platform on Gemini models, connecting to Google Workspace, Microsoft 365 and business data); Amazon Bedrock (AWS) (AWS service giving one API to foundation models from Amazon, Anthropic, OpenAI, Meta and others, plus AgentCore for running agents); Microsoft Foundry (formerly Azure AI Foundry) (Azure platform for building AI apps and agents on models from OpenAI, Anthropic, Meta and others; includes Azure OpenAI); GitHub Copilot (AI coding assistant and coding agent in the IDE, CLI and GitHub.com, with Business and Enterprise plans billed per seat plus AI credits). Each has its own page with products, pricing, ownership and the banks among the 120 largest US banks that are named as customers. More vendors in this category are being added as their pages are researched and sourced.
When a bank picks an enterprise AI assistants vendor, you'll read it here first.
when a bank picks an AI vendor, or a vendor ships for banks, it is in the next morning's brief · six sourced stories · 7 am PT · free
plus every regulation, bank and agent page update, the morning after · leave any morning