Selection guide
ChatGPT Enterprise vs Claude Enterprise: pricing and how to choose
Enterprise AI assistant plan. A company subscription to a vendor's AI assistant, such as ChatGPT Enterprise or Claude Enterprise, that adds central billing, identity and admin controls and business data terms to the product staff use day to day.
1AYM is an OpenAI Select Partner.
Checked . Prices were read from the UK, signed out, on each vendor's own pricing page, so they are the figures a UK buyer sees; plan details come from the vendors' pricing and help pages. Anthropic states that its pound prices include 20% VAT. OpenAI's page does not say whether its pound prices include VAT. Anthropic shows Claude Enterprise in US dollars, even to UK visitors. Vendors change prices and plans without notice, so confirm them on the sources below before you buy. The US-dollar section and the audit log table were read on 30 September 2026.
The four plans side by side
What each vendor publishes for its team plan and its enterprise plan. The numbers in brackets point to the sources at the foot of the page. Where a vendor publishes nothing, the cell says so rather than filling the gap.
| Dimension | Claude Team | Claude Enterprise | ChatGPT Business | ChatGPT Enterprise |
|---|---|---|---|---|
| UK price, billed annually [1, 3, 14] | £18 per user a month for a standard seat; £90 for a premium seat. Includes 20% VAT | US$20 per seat a month, plus all usage at Anthropic's API rates. Shown in US dollars to UK buyers | £15 per user a month for a standard seat; £75 for a premium seat. OpenAI does not say whether VAT is included | Quote-only: custom pricing through OpenAI's sales team, credit-based or token-based |
| UK price, billed monthly [1, 14] | £22.50 standard; £113 premium. Includes 20% VAT | Not offered: annual billing only | £18 standard; £90 premium. OpenAI does not say whether VAT is included | Not offered: annual plans |
| US list price, for reference only: annual or monthly [1, 9, 10] | US$20 or US$25 standard; US$100 or US$125 premium | US$20 per seat, annual only, plus usage | US$20 or US$25 standard; US$100 or US$125 premium | Quote-only |
| Seats [1, 2, 3, 5, 10, 14] | 2 to 150 | At least 20 bought online; at least 50 through sales | At least 2; OpenAI describes it for teams of 2 to 200 | Not published |
| Smallest annual commitment, seats only (our arithmetic) [1, 5, 14] | £432 (2 standard seats at £18 a month), VAT included | US$4,800 (20 seats at US$20 a month), before any usage | £360 (2 standard seats at £15 a month), VAT basis not stated | Not published |
| How usage is limited [1, 3, 10, 13] | An allowance per seat that resets every five hours, plus weekly limits; extra usage can be bought | No per-seat limit: every token is billed, and admins set spend limits | Included limits per seat; premium seats have no five-hour limit; workspace credits extend usage | Set by contract: credit-based pricing with no fixed rate limits, or fixed per-seat limits |
| Coding agent access [1, 3, 9] | Claude Code, drawing on the seat's usage limits | Claude Code, with its usage billed at API rates on top of the seat | Codex | Codex |
| Identity [1, 9] | Single sign-on (SSO) and domain verification; no SCIM (automatic user sync from your identity provider) | SSO, SCIM, role-based access and domain capture (accounts on your email domain join the company workspace) | SAML SSO, MFA and domain verification; no SCIM | SAML SSO, SCIM provisioning and role-based access controls |
| Audit and records [1, 3, 9] | No audit logs, Compliance API (lets your compliance tools pull activity and conversation records) or custom retention | Audit logs, Compliance API and custom data retention | No Compliance API | Compliance API and custom data retention |
| Encryption keys and network [1, 3, 9] | Not listed for this plan | Customer-managed encryption keys (you hold the keys to your stored data) and IP allowlisting (access only from your company's network addresses) | Not included | Enterprise Key Management (OpenAI's customer-managed keys) and IP allowlisting |
| Where data is held [3, 4, 7, 9, 12] | Stored in the US; processing may be routed to the US, Europe, Asia or Australia | As Team, with an option to keep inference (the model's processing of each request) in the US at 1.1 times API rates on newer models | No data residency option | A choice of ten storage regions, including the UK; in-region inference only in Europe, the US and the UAE |
| Training on your content [1, 9] | Not by default | Not by default | Not by default | Not by default |
| Workplace connectors [2, 3, 9] | Google Drive, Gmail, Google Calendar, GitHub, Microsoft 365 and Slack | Google Drive, Gmail, Google Calendar, GitHub, Microsoft 365 and Slack | Google Workspace, Slack, GitHub, Microsoft 365 and more | As Business, plus a connector registry for admins |
| Context window: how much text it can take in at once [6, 9] | 200K tokens by default; up to 1M on named models | 200K tokens by default; up to 1M on named models | 54K on GPT Instant; 256K on GPT Reasoning | 128K on GPT Instant; 256K on GPT Reasoning |
| Paying [1, 4, 9, 11] | Card | Card or ACH (US bank transfer) online, in US dollars only; invoicing and other currencies through sales | Card only; no invoicing or purchase orders | Invoicing and volume discounts |
ChatGPT Enterprise pricing in US dollars, beside the other three plans
ChatGPT Enterprise has no published price in US dollars or in any other currency. OpenAI sells it by quote, and bills usage either against its credit rate card or, under a token-based agreement, in US dollars against its Enterprise rate card and your contract [9, 17]. For scale, ChatGPT Business and a standard Claude Team seat both list at US$20 a user a month billed annually [15, 17].
| Dimension | Claude Team | Claude Enterprise | ChatGPT Business | ChatGPT Enterprise |
|---|---|---|---|---|
| Seat, billed annually [9, 15, 17] | US$20 per user a month for a standard seat; US$100 for a premium seat | US$20 per seat a month, plus all usage at Anthropic's API rates | US$20 per user a month for a standard seat; US$100 for a premium seat | Quote-only |
| Seat, billed monthly [9, 10, 15, 17] | US$25 standard; US$125 premium | Not offered: annual billing only | US$25 standard; US$125 premium | Not offered: annual plans |
| Usage beyond the seat [1, 3, 16, 17] | An allowance per seat; extra usage can be bought | Every token billed at API rates: US$2 per million tokens in and US$10 out on Claude Sonnet 5.5; US$4 and US$20 on Claude Opus 5.5 | Included limits per seat, then workspace credits, charged per model in credits per million tokens | By agreement: OpenAI's credit rate card, or usage billed in US dollars against its Enterprise rate card |
| Smallest annual commitment, seats only (our arithmetic) [1, 5, 15, 17] | US$480 (2 standard seats at US$20 a month) | US$4,800 (20 seats at US$20 a month), before any usage | US$480 (2 standard seats at US$20 a month) | Not published |
Anthropic's US-dollar prices exclude applicable tax [15]; OpenAI's pricing documentation does not state a tax basis.
The per-token rates are why a Claude Enterprise seat price says little about the bill. A month of 2 million tokens read and 200,000 written comes to US$6 of usage on Claude Sonnet 5.5, or US$12 on Claude Opus 5.5, on top of the US$20 seat (our arithmetic on Anthropic's list rates) [16]. The only forecast worth signing against is your own pilot's usage per person.
On ChatGPT Enterprise the question is which rate card your agreement uses. OpenAI's credit rate card spreads a hundredfold by model: GPT-6 Astra is 250 credits per million input tokens and GPT-6 Luna is 2.5 [17]. The model your staff default to moves the bill as much as the seat count does, so get the US-dollar price of a credit, the rate card that applies and what happens when credits run out in writing before you sign. Setting caps and charging usage back to teams is the subject of our guide to forecasting and capping AI spend.
Which plan fits which situation
The seat price separates these four plans less than it appears to. Two of them are priced within a few pounds of each other, and the other two charge in ways that make a seat price the wrong number to compare. Find the situation below that matches yours.
- Up to 150 people, and a fixed bill matters
- Claude Team or ChatGPT Business. A UK buyer sees £18 and £15 a standard seat a month billed annually, and Anthropic's figure includes 20% VAT while OpenAI's page does not say [1, 14]. Once VAT is settled on your order form the gap is small, so the deciding question is which assistant your people get more done with, and only a pilot on your own work answers it. Claude Team is for teams of 2 to 150; OpenAI describes ChatGPT Business for teams of 2 to 200 [1, 14].
- Security requires SCIM, audit logs or a compliance API
- That means an enterprise plan on either side. Neither team plan includes SCIM provisioning or a Compliance API [1, 9], so a security review that requires them has already ruled both out.
- Data must be stored in the UK
- On the published terms, ChatGPT Enterprise offers UK storage at rest for new workspaces [12]. Anthropic states that data on its commercial products is stored in the US, and its Enterprise option keeps inference in the US rather than the UK [3, 7]. Anthropic also offers regional residency for Claude models through AWS, Google Cloud and Microsoft, but its regional compliance page lists Europe, the US, Canada and Asia-Pacific and names no UK region [8], and that route is a build on those platforms rather than the Claude app.
- Usage varies widely from person to person
- Claude Enterprise bills every token at API rates on top of the US$20 seat [3], so light users cost little more than the seat and heavy users cost what they consume. The team plans handle the same spread with premium seats at five times the usage [1, 10], which suits a small, known group of heavy users.
- You need to pay by invoice, or in pounds on an enterprise plan
- ChatGPT Business is card-only [11], and Claude Enterprise bought online is billed in US dollars only [4], though Claude Team is priced in pounds for UK buyers [1]. Invoicing, and an enterprise plan in other currencies, mean a sales-led contract on either side: Claude Enterprise through Anthropic's sales team, which starts at 50 seats [3], or ChatGPT Enterprise [9].
- Engineers are a large share of the users
- All four plans give access to a coding agent: Claude Code on the Claude plans and Codex on the ChatGPT plans [1, 9]. On Claude Enterprise its usage is billed at API rates on top of the seat [3, 4]. The licence is the smaller half of that decision. The larger half is the operating layer described in our guide to rolling out Codex and Claude Code to engineers.
How the two enterprise plans charge
Claude Enterprise is a seat plus usage. The seat is US$20 per user a month, billed annually, and Anthropic shows it in US dollars even to UK buyers [1]. It covers access only; every token your staff use, in the Claude apps or in Claude Code, is billed at Anthropic's standard API rates on top [3, 4]. Bought online it needs at least 20 seats, which is US$4,800 a year before any usage (our arithmetic), and usage is prepaid as credits: when the balance reaches zero, usage stops for the whole organisation until an owner buys more [4, 5]. Bought through Anthropic's sales team it needs at least 50 seats, and usage is invoiced monthly in arrears [3, 4].
ChatGPT Enterprise has no published price. OpenAI's pricing page lists custom pricing through its sales team, with credit-based and token-based options [9]. Treat any figure you find elsewhere as someone else's estimate, not a price you can budget against.
Two contract terms move the bill more than the seat price does. Seats added to an annual ChatGPT Business plan stay committed to the end of the term [11], and seats on Claude Enterprise bought online cannot be removed mid-term [5]. Existing Claude Enterprise customers on the older Standard and Premium seats move to the single usage-billed seat at their next renewal [4], so that renewal is the moment to re-cost the plan.
Audit logs and the Compliance API: what each enterprise plan records
Both enterprise plans have a Compliance API, and neither team plan has one [9, 15]. The two differ most in how long records are kept and in what they capture from agents working on a user's own machine. Claude Enterprise keeps six years of activity in its feed, recorded from the day the API is switched on [20]. OpenAI's Compliance Logs Platform keeps records for 30 days, so you export them continuously to your own systems [22].
| Dimension | Claude Enterprise | ChatGPT Enterprise |
|---|---|---|
| Which plans have it [9, 15, 22] | Enterprise only; Claude Team has no audit logs and no Compliance API | Eligible Enterprise and Edu workspaces; ChatGPT Business has no Compliance API |
| Admin and sign-in records [18, 21] | An audit log export (CSV) covering the past 180 days: sign-ins, member and SSO changes, project changes and file uploads. Chat and project titles and content are left out; only their IDs are exported | Authentication logs are one of the Compliance API's record types |
| Conversation content [20, 21, 22] | Message bodies and file contents through the API's content endpoints; the activity feed records who did what and when, without prompt text | Conversation messages; on ChatGPT Work, user prompts and agent responses |
| Coding and desktop agents [20, 21, 22] | Transcripts of Cowork, Claude Code and Claude for Microsoft 365 sessions run under the Enterprise account. Claude Code on a Console API key, through a third-party cloud or as a cloud session is not captured | Codex usage logs and ChatGPT Work prompts and responses. OpenAI states these are not a complete record of every file operation, shell command, browser action, tool call or approval |
| How long records are kept [20, 22] | Activity feed: six years, recorded from the day the API is switched on, with no backfill. Content follows your retention setting | 30 days on the Compliance Logs Platform; for longer, export continuously to an eDiscovery, data loss prevention (DLP) or security information and event management (SIEM) system |
| Access [20, 21] | The primary owner switches it on in organisation settings, and it covers every linked organisation | An Enterprise Compliance API key, used within your admin role's permissions |
Neither vendor presents its Compliance API as a full record of what an agent did on a device [20, 23]. A security review that needs that record plans for OpenTelemetry or endpoint logging as well, and sets the export schedule before the first user signs in.
What to test in a pilot
A pilot answers the questions a pricing page cannot. Run both vendors on the same real work for two weeks, with a named owner and one question to settle, and collect evidence on each point below.
- Real work, not demonstrations
- Pick two or three tasks your teams do every week, use the documents they really work from, and have the same people try both assistants on them.
- The usage bill per person
- On Claude Enterprise, read the spend per user from the pilot and project it across a year before signing, because usage has no cap unless an admin sets one [3]. On the team plans, count how many people reach their limits, because each of them is a premium seat [1, 10].
- Your identity provider
- Connect single sign-on, and SCIM where the plan has it, to the directory you actually run, and check that joiners and leavers flow through without manual work. Test leavers hardest, because nobody reports access that should have gone. Our identity provisioning for 1,000+ users at a large international marketing agency ends Claude access when the HR record changes, not when a colleague remembers.
- Your own documents and mail
- Connect the Microsoft 365 or Google Workspace estate your staff use, and check what each assistant can see and whose access rights it follows.
- Residency and retention in writing
- Put the data region, the retention setting and the training position into the order form, not only the pricing page. Anthropic states that its prices and plans can change at its discretion [1].
What a switch costs beyond the seats
A seat price comparison leaves out most of what a switch costs. None of the items below appears on either vendor's pricing page, and each should be costed before a switch is approved.
- Overlapping contracts
- Both vendors sell annual terms [4, 11], so a switch before renewal means paying for both plans until the old term ends.
- Rebuilt workspace content
- Plan for someone to rebuild and re-test custom GPTs, Claude projects and shared instructions in the new product, and ask the vendor what, if anything, it imports. Our guide to one-line rules for ChatGPT and Claude projects shows where each product keeps project instructions.
- A second security review
- Every connector is authorised again, and the review your security team completed for the first vendor is repeated for the second.
- Records you must keep
- Conversations held under the old vendor's retention settings stay with that vendor, so decide how you will meet any retention or disclosure duty before the account closes.
- Coding agent setup
- Repository rules, skills and permissions written for one coding agent do not carry to the other unless they were built to. That is the gap 1AMS, our product for building skills once and deploying them to each vendor's tools, is designed to close. For the instruction file itself, our guide to writing AGENTS.md and CLAUDE.md shows how one file can serve both agents.
Where 1AYM stands
1AYM is an OpenAI Select Partner, and the Claude certifications on this site belong to our founder personally, verifiable on Credly. We build on both vendors' products for clients, so this page matches a plan to a situation and never to a vendor. What our OpenAI partner status covers is set out on its own page.
If the open question is whether an assistant plan is the right investment at all, that is the build-versus-buy decision our AI strategy work settles before anyone buys seats.
For engineers and security reviewers: controls to check
The table shows what each plan includes. These are the details behind it that a security review asks about, each with its source.
- SCIM and seat assignment
- SCIM is on both enterprise plans and on neither team plan [1, 9]. On Claude Enterprise, users provisioned by JIT or SCIM take a seat automatically, and provisioning does not complete when no seat is free [5].
- Compliance API scope
- Anthropic's Compliance API returns activity logs, chat histories and file content, filterable by user and time range [3]. OpenAI lists audit logs and usage monitoring through its Compliance API on Enterprise [13].
- Compliance API keys and limits on Claude
- A Compliance Access Key created in claude.ai reaches every endpoint; an Admin API key from Claude Console reaches the activity feed only. All compliance endpoints share a limit of 600 requests a minute per parent organisation [19].
- Deletes through the API are final
- Claude's Compliance API can delete chats, files and projects, and those deletes are immediate and cannot be recovered. Content a user deletes in claude.ai is not recoverable either, so pull anything under legal hold while it still exists [19, 20].
- What happens on the device
- Anthropic captures Cowork and Claude Code sessions as their requests reach its API, so on-device activity that never reaches the API is not in the transcript; both tools can also stream OpenTelemetry events to a collector you run [19, 20]. OpenAI says local execution under ChatGPT Work produces separate OpenTelemetry events, and that conversation records alone do not make a complete record of local actions [23].
- Audit log export with your own keys
- A Claude organisation that uses customer-managed encryption keys cannot use the audit log export button; the same events are available through the Compliance API [18].
- Encryption keys
- Claude Enterprise supports customer-managed encryption keys provisioned in your own cloud provider [3]. ChatGPT Enterprise lists Enterprise Key Management [9].
- Storage residency is not inference residency
- OpenAI's UK option covers storage at rest of conversations, files, memory, custom GPTs and generated images. Inference residency is offered only in Europe, the US and the UAE, and authentication, routing and data sent through external integrations may be processed outside the region [12].
- US-only inference on Claude
- Claude Enterprise organisations can keep inference in the US; usage is then billed at 1.1 times standard API rates for Claude Opus 4.6, Claude Sonnet 4.6 and later models [4].
- Domain capture and role-based access on Claude Team
- Anthropic's sources disagree. The pricing table marks both as not included on Team [1]; the Team help article lists Domain Capture and role-based permissioning as Team features [2]. The table above follows the pricing page. Ask Anthropic which is current before a security review relies on either.
- Network controls
- IP allowlisting is listed on both enterprise plans [1, 9]. Claude Enterprise also lists network-level access control [1].
Sources
- [1]Anthropic, Claude plans and pricing (Team and Enterprise), as shown to a UK visitor, read 29 September 2026
- [2]Anthropic Help Centre, What is the Team plan?, read 29 September 2026
- [3]Anthropic Help Centre, What is the Enterprise plan?, read 29 September 2026
- [4]Anthropic Help Centre, How am I billed for my Enterprise plan?, read 29 September 2026
- [5]Anthropic Help Centre, Purchase and manage seats on Enterprise plans, read 29 September 2026
- [6]Anthropic Help Centre, How large is the context window on paid Claude plans?, read 29 September 2026
- [7]Anthropic Privacy Center, Where are your servers located? (dated 15 June 2026), read 29 September 2026
- [8]Anthropic, Regional compliance and data residency, read 29 September 2026
- [9]OpenAI, Business pricing (ChatGPT Business and Enterprise), read 29 September 2026
- [10]OpenAI Help Center, ChatGPT Business overview, read 29 September 2026
- [11]OpenAI Help Center, Managing billing and seats in ChatGPT Business, read 29 September 2026
- [12]OpenAI Help Center, Data residency and inference residency for ChatGPT, read 29 September 2026
- [13]OpenAI, ChatGPT pricing and usage limits (learn.chatgpt.com), read 29 September 2026
- [14]OpenAI, ChatGPT pricing (Business and Enterprise), as shown to a UK visitor, read 29 September 2026
- [15]Anthropic, Claude plans and pricing, US-dollar view (prices exclude applicable tax), read 30 September 2026
- [16]Anthropic, Claude API pricing, the rates Claude Enterprise usage is billed at, read 30 September 2026
- [17]OpenAI, ChatGPT pricing and credit rate card (learn.chatgpt.com), US-dollar view, read 30 September 2026
- [18]Anthropic Help Centre, Access audit logs (updated 15 June 2026), read 30 September 2026
- [19]Anthropic, Compliance API (Claude Platform Docs), read 30 September 2026
- [20]Anthropic, Compliance API FAQ (Claude Platform Docs), read 30 September 2026
- [21]OpenAI, Compliance API and audit events (ChatGPT Learn), read 30 September 2026
- [22]OpenAI, ChatGPT Work admin FAQ (ChatGPT Learn), read 30 September 2026
- [23]OpenAI, Governance (ChatGPT Learn), read 30 September 2026
Pricing questions
How much does Claude Enterprise cost?
Anthropic lists Claude Enterprise at US$20 per seat per month, billed annually, plus all usage at its standard API rates, and shows that price in US dollars even to UK buyers. The seat covers access only and includes no usage. Bought online the minimum is 20 seats; through Anthropic's sales team it is 50. Checked on Anthropic's pricing and help pages on 29 September 2026.
How much does ChatGPT Enterprise cost?
OpenAI does not publish a price, in US dollars or any other currency. Its pricing page lists custom pricing through the sales team, with credit-based and token-based options; under a token-based agreement, usage is billed in US dollars against OpenAI's Enterprise rate card and your contract. Any figure quoted elsewhere is an estimate, not a price from OpenAI. For scale, ChatGPT Business lists at US$20 a user a month billed annually. Checked on OpenAI's pricing page on 29 September 2026 and its rate card documentation on 30 September 2026.
How much is ChatGPT Business?
For a UK buyer, a standard seat is £15 per user per month billed annually or £18 monthly, and a premium seat with five times the usage is £75 or £90. OpenAI's page does not say whether these include VAT. A workspace needs at least two seats, and OpenAI describes the plan for teams of 2 to 200. Checked from the UK on 29 September 2026.
Is Claude Team the same price as ChatGPT Business?
Not quite, and the pound figures are not like for like. A UK buyer sees Claude Team at £18 a standard seat a month billed annually and ChatGPT Business at £15. Anthropic's figure includes 20% VAT, which puts it at £15 before VAT (our arithmetic); OpenAI's page does not say whether its figure includes VAT, so confirm that before comparing. The plans also differ in size (Claude Team is for 2 to 150 people, ChatGPT Business for 2 to 200), in context window and in how usage limits work, which a pilot on your own work will show.
Can we keep our data in the UK?
On the published terms, ChatGPT Enterprise offers UK storage at rest for new workspaces, though in-region inference is available only in Europe, the US and the UAE. Anthropic states that data on its commercial products is stored in the US, with an option on Claude Enterprise to keep inference in the US. Neither team plan offers a residency choice.
Does either vendor train its models on our data?
Both say no by default on these plans. Anthropic lists no model training on your content by default, and OpenAI lists no training on your business data by default.
Does ChatGPT Enterprise have audit logs?
Yes, through OpenAI's Compliance API, which covers conversation messages, Codex usage and authentication logs, and ChatGPT Work prompts and responses on eligible Enterprise and Edu workspaces. The Compliance Logs Platform keeps records for 30 days, so export them continuously to your own eDiscovery, data loss prevention or security monitoring (SIEM) system. ChatGPT Business has no Compliance API. Checked on OpenAI's documentation on 30 September 2026.
What does the Claude Compliance API capture?
On Claude Enterprise it returns an activity feed of who did what and when, kept for six years, starting from the day the API is switched on, plus chat, file and project content and transcripts of Cowork and Claude Code sessions run under the Enterprise account. Owners can also export a CSV audit log covering the past 180 days. Claude Team has neither. Checked on Anthropic's documentation on 30 September 2026.
Are the prices in pounds, and do they include VAT?
The team plans are shown in pounds to a UK buyer. Anthropic states that its Claude Team prices include 20% VAT. OpenAI's pricing page does not say whether its ChatGPT Business prices in pounds include VAT, so ask before you compare the two. Claude Enterprise is shown in US dollars even to UK buyers, and bought online it is billed in US dollars only; ChatGPT Enterprise is priced by quote. Check the figure at checkout or on your order form.
In this topic
Which AI assistant should we buy, and how do we roll it out?
What is Claude Cowork, and does it matter to my company?
- ChatGPT Work vs Claude CoworkComparison guide
Should we roll out ChatGPT Work or Claude Cowork to our staff?
- ChatGPT Enterprise rolloutRollout guide
How do we roll out ChatGPT Enterprise, and in what order?
- Microsoft 365 Copilot vs ChatGPT EnterpriseComparison guide
Our firm runs on Microsoft 365. Should we buy Microsoft 365 Copilot or ChatGPT Enterprise?
What should we fix for oversharing before we roll out Microsoft 365 Copilot?
How do I write custom instructions so ChatGPT or Claude stops explaining my own job to me?
How do I stop ChatGPT or Claude making the same mistake after I have corrected it?
How do I turn my review comments into an AI checklist my team runs before I see their drafts?
How do you get AI to follow your brand guidelines?
How do I make AI write like me?
How do I use a meeting transcript to get honest feedback on how I ran the meeting?
How do we share AI skills across a company in Claude and ChatGPT?
How do you stop AI inventing facts about your company?
How do I turn a chat into a reusable skill in Claude or ChatGPT?
Further
- AI governance implementation · Identity, logging and permissions built as controls rather than written as policy.
- Engagement file D-01 · The documented case: the platform engineering underneath an agency's organisation-wide AI programme.
- Rolling out Codex and Claude Code · The operating layer around the coding agents all four plans include.
We build these systems for a living. See the engagement files for what that looks like in practice, or write to us if yours is the next one.
Last reviewed · 1AYM