Claude Enablement: Claude Code and Cowork set up for your team in three days
Claude Enablement is a fixed-price, on-site install of a working Claude workspace for a UK firm: the Claude Team or Enterprise tenancy configured, connectors to your systems on least-privilege scopes, context files and role prompts written for your people, shared skills built from your real work, and the governance settings recorded. Non-technical staff work in Claude Cowork. The founder, operator or technical lead gets Claude Code. Every seat is trained on their own tasks, and we stay for 30 days afterwards, inside the price, because that is when adoption is won or lost.
Two packages: Team Install from GBP 4,500 ex VAT for a founder and up to five seats, and Business Install from GBP 7,500 ex VAT for up to twelve. Delivered by The AI Consultancy, an Anthropic Consulting Partner with Claude Certified Architect practitioners, from our Hoxton office at 20 Wenlock Road, London, N1 7GU, on site anywhere in the UK.
This is the enablement tier of our Claude work. If you need a system built and integrated, see Claude Implementation. If you need advice before deciding, see Claude Consulting.
What is installed
A Claude workspace is a stack of small, owned configurations rather than one clever prompt. The install leaves each layer in place, written down, with a named owner on your side.
- Tenancy. Your Claude Team or Enterprise organisation, with a named client admin, seat types matched to each role, single sign-on where your identity provider supports it, and the plan-level data settings checked and recorded.
- Connectors.The systems your people actually use (Google Workspace or Microsoft 365, your CRM, your finance platform, your file store) connected through Anthropic's connectors or the Model Context Protocol, enabled at organisation level, authenticated per person, and scoped to the least access the use case needs. Write actions are set to need approval unless you decide otherwise.
- Context files. One organisation instruction carrying who you are, your tone and your data boundaries, plus a project instruction and a seeded memory for each recurring workflow, so Claude is useful from the first prompt rather than after a month of correction.
- Skills. A small, controlled set of skills built from the procedures your team repeats: the monthly report, the tender qualification, the client follow-up. Each has an owner, a risk tier and a test case, and each is provisioned to the seats that need it.
- Role prompts. One prompt per role, not per person, under 500 words, covering role context, output format, domain knowledge to apply and the things that must never be automated. Adding a seat later means assigning a role prompt, not writing a new one.
- Governance settings. The Cowork permission modes, connector tool approvals, web search and network settings as configured, plus the data exclusions list and the human-sign-off rules for anything that sends, pays, signs or updates a production record.
What is deliberately not included
The fixed price holds because the scope is fixed. Four things sit outside it and are quoted separately, each through the service line already built for it.
- Custom connectors, skills and applications. A bespoke MCP server for a line-of-business system, an evaluation harness or a custom application is AI Implementation work.
- Dashboards and agents. Autonomous agents, operations dashboards and customer-facing automation belong to Agentic AI and Workflow Automation.
- Enterprise tenancy configuration. SCIM provisioning, role-based access design, audit-log integration with your SIEM and custom retention policies are scoped as part of Claude Implementation, usually alongside an Enterprise purchase.
- Licences. Claude seats are bought by you from Anthropic. We help you choose the plan and seat mix; we do not resell.
Who it is for
UK firms of roughly five to one hundred staff where AI use today is either blocked by policy or confined to a few individuals on personal accounts with no shared setup, no connectors and nobody accountable for what goes in. In practice that describes most owner-managed businesses and professional practices we meet: recruitment and staffing firms, accountancy and legal practices, security and facilities businesses, manufacturers and producers, and specialist finance houses.
The common failure we are asked to fix is not a missing tool. It is a founder who bought licences, watched two people use them for a fortnight, and now has a subscription nobody opens. The public version of the same failure is the founder-terminal install: a day burned collecting credentials and exporting chat history, a founder handed a terminal and a training burden they cannot carry, and an installer who leaves with no follow-up. Our method removes each of those: the questionnaire and gap analysis happen before we arrive, so the on-site days are spent on your work rather than on API keys; Cowork is the default surface, so non-technical seats never see a terminal; and the 30-day adoption window is inside the price, so the hand-off is a plan rather than a goodbye.
It is not for firms that need a bespoke build, an API integration or a regulated-sector evaluation programme; those start with Claude Implementation. And it is not for firms whose client data cannot be processed by a cloud model under any agreement; those start with Private AI Concierge.
The three days, and the 30 days after
The on-site days only work because of what happens before them. A structured questionnaire built from the scoping call captures your roles, workflows, systems, data sensitivity and IT position; we run a gap analysis on the return and do not book build day until the critical gaps are closed. Business Install adds a third on-site day and a second follow-up session; the shape is otherwise the same.
Before we arrive: questionnaire and gap analysis
Named users and roles, top three workflows per role with real examples, output formats, data classification, the systems to connect and who owns each, IT sign-off status, and the documents approved for Claude to see. We draft the organisation instruction, the role prompts and the project seeds from this before day one, so the site days are for testing against real work rather than starting from a blank page.
Day one: tenancy, connectors, governance
The Team or Enterprise organisation is configured with your admin, seats are assigned, single sign-on is set up where available, the plan-level data settings are checked, and each connector is enabled at organisation level and scoped with your IT lead in the room. Write actions are set to need approval. Cowork permissions, web search and network settings are configured and written into the governance record.
Day two: seats, role prompts, skills, live examples
Each person is set up in turn: role prompt applied, projects created and seeded, skills provisioned, connectors authenticated under their own permissions, and one of their real tasks completed with them at the keyboard. Non-technical seats work in Cowork and chat. The operator or technical lead is trained on Claude Code and on maintaining the context files and skills library.
Follow-up day, remote: what broke, what stuck
About a week later. Each seat brings the tasks that worked and the ones that did not; prompts, projects and skills are adjusted; the usage analytics your plan provides are read against the adoption tracker; and the acceptance checklist is signed off. The handover pack is issued in writing.
Days 8 to 30: the adoption window
Inside the fixed price. A named consultant is available for questions by email and a weekly check-in, the adoption tracker is reviewed each week, and at day 30 you get an honest keep-fix-retire review of every workflow, skill and connector. In our own delivery records, the installs that lost momentum did so after the on-site days, once the initial energy faded and nobody owned the follow-up. This window exists to stop that.
Two packages, fixed price
Both prices are ex VAT and follow our published engagement terms: card via Stripe under GBP 5,000, otherwise staged invoicing on 14-day terms. UK travel is itemised separately only where the engagement needs sustained on-site delivery beyond the days below.
| Included | Team Install | Business Install |
|---|---|---|
| Price | From GBP 4,500 ex VAT | From GBP 7,500 ex VAT |
| Seats set up | Founder plus up to five | Up to twelve |
| On-site days | Two | Three |
| Remote follow-up | One day | Two sessions |
| Support and adoption window | 30 days | 30 days |
| Questionnaire and gap analysis | Included | Included |
| Tenancy, connectors, governance record | Included | Included |
| Role prompts, project seeds, skills | Per role, typically two to three roles | Per role, typically four to six roles |
| Claude Code for the operator | One person | One to two people |
| Handover pack | Included | Included |
| Additional seats | GBP 250 to 350 each, including a role prompt and one onboarding session | |
After the 30 days
Most clients need nothing further. Where a firm wants continuity, an optional monthly follow-on at GBP 750 to 1,500 covers a usage review, a skills refresh and a quarterly check-in, and is the natural route into our Fractional Chief AI Officer retainer for firms that want ongoing AI leadership. If the install surfaces a build that needs doing, 50 per cent of the install fee is credited against a Discovery and Pilot engagement commissioned within 90 days, on the same terms as the Readiness Sprint on our pricing page.
Team or Enterprise: what each plan actually gives you
We install on whichever plan fits, and we say plainly what each one can and cannot do. The rows below were checked against the UK render of claude.com/pricing and Anthropic's Help Center on 15 September 2026; prices are Anthropic's, paid by you, and the page states that they include 20 per cent VAT.
| Question | Claude Team | Claude Enterprise |
|---|---|---|
| Seats and price | 2 to 150 seats. Standard £18 a seat a month on annual billing (£22.50 monthly); premium £90 (£113). Seat types can be mixed. | £18 a seat a month plus usage billed at API rates; annual billing; seat minimums apply. |
| Cowork and Claude Code | Both included. Cowork on by default; cloud sessions on by default, switchable by an Owner. | Both included. Cloud sessions off by default and granted per group through custom roles. |
| Identity and admin | Single sign-on, domain verification, central billing, spend controls, organisation instructions, admin controls for connectors, organisation-wide skills deployment. | All of Team plus SCIM provisioning, role-based access, domain capture and custom data retention controls. |
| Monitoring across the organisation | Admin usage analytics: weekly active members, adoption by product, connector read and write counts, skills usage, spend export. Cowork events can be streamed to a SIEM via OpenTelemetry. No audit logs. | Usage analytics plus audit logs and the Compliance API, which captures Cowork sessions on web, desktop and mobile. |
| Model training on your content | None by default. | None by default. |
| Context window | 200k, as listed on the plan page. | 500k on the default model, as listed on the plan page. |
The individual plans (Free, Pro and Max) have no organisation-wide administration, no usage analytics and a model-improvement setting that is on unless someone turns it off. We do not install a team on them. For the full comparison, including the seat minimums on each Enterprise purchase route, see our guide to Claude Team versus Enterprise.
Governance: UK GDPR, data settings, connector scopes, and when IT says no
UK GDPR regulates your processing of personal data, not the software. On Team and Enterprise Anthropic acts as processor under its commercial terms and Data Processing Addendum, and you remain controller. The install therefore records the things a controller has to be able to show: which data classes may be used in which Claude surfaces, the lawful basis for each workflow that touches personal data, whether a DPIA screen is triggered, and the retention position of the plan you are on. Our guides to whether Claude is GDPR compliant and whether you need a DPIA cover the reasoning.
- Data-training settings. Team and Enterprise do not train on your content by default. The individual plans carry a model-improvement setting, and Anthropic states that when it is on, chats may be retained in de-identified form for up to five years. We check the setting on every account that touches real data and record the position in the handover pack.
- Connector scopes. On Team and Enterprise an Owner enables each connector for the organisation, every person authenticates individually, and Claude can only see what that person can already see in the source system. Owners can then restrict write tools to Needs approval or Blocked for the whole organisation. We start read-only, add write access only where a workflow needs it and the sponsor has signed, and keep a connector inventory that lists every scope.
- Connected services are still third parties. Xero, HubSpot, Google and Microsoft process the data on their own infrastructure under their own terms, and Anthropic's plan settings do not change where they run. The inventory names each of them so your Article 30 record is complete.
- Human sign-off. Nothing in the install lets Claude send client email, submit filings, make payments, sign documents, give regulated advice or update a production finance, CRM or case record without a person approving it. Those rules are written into the organisation instruction and the connector settings, not left to memory.
When IT says no
The objection is usually to the version of AI use that is already happening: staff on personal accounts, client data pasted into chats, no record of any of it. A governance-first pilot answers it with a company-owned tenancy, a defined pilot group, approved use cases, a written list of banned data types, read-only connectors, and a review date with usage evidence. Where the IT lead needs audit logs, custom retention or SCIM before they will sign, that is an Enterprise requirement and we say so up front rather than promising Team can do it. Where a regulator or client contract rules out cloud processing altogether, we do not install Claude and point you to Private AI Concierge instead.
Delivery evidence
The two packages are the published form of installs we have been delivering since June 2026, on Pro, Team and Enterprise tenancies, for firms in pharma recruitment, engineering staffing, professional services, wine production, security, resource finance and paper products. Typical installs cover five to twelve seats over two or three on-site days with a remote follow-up, which is why the packages are shaped that way. We do not publish client names or outcome figures for these engagements without written consent, and we do not invent them; case studies are added only when a client has approved one.
What we can say from the records is where the work is won. Where an install lost momentum, it did so after the on-site days, when some seats were active and others were not, and nobody on either side owned the follow-up. The 30-day window, the role-prompt-per-role rule and the questionnaire-first sequence are direct responses to that pattern.
Anthropic Consulting Partner
Claude delivery with Claude Certified Architect practitioners
Method, not improvisation
Questionnaire, gap analysis, role prompts, skills register, 30-day plan
Governance first
Read-only by default, write actions need approval, every scope recorded
UK delivery
London-based, on site anywhere in the UK
Frequently asked questions
Can our staff use Claude Code if nobody here can write code?+
Is Claude Code worth it for a small business owner, or is the web app enough?+
Can we get Claude Team for two users?+
How much does Claude cost a small business on top of the install?+
Our IT lead says no. Who is right?+
Is it safe to connect Claude to Xero or HubSpot?+
Will Anthropic train on our data?+
Can you monitor how the team is using Claude across the organisation?+
What do we get at handover?+
Anthropic's Claude for Small Business toggle is free. Why pay for an install?+
Related Claude services and reading
- Claude Implementation: the build tier. API integration, bespoke connectors, evaluation harnesses and Enterprise tenancy design, scoped after a free consultation.
- Claude Consulting: the advisory tier. Readiness, use-case prioritisation, plan selection and ongoing oversight for firms deciding whether and where Claude fits.
- Setting up Claude for a business: the full stack: the layers an install leaves in place, and who owns each.
- A 30-day Claude adoption plan for UK teams: the week-by-week shape of the adoption window.
- How much does Claude cost in the UK?: Anthropic's sterling seat prices with the working shown.
Readiness Sprint from GBP 3,500 · Claude Enablement from GBP 4,500 · Discovery and Pilot from GBP 15,000 · Build and Embed from GBP 40,000 · Day rate GBP 950 to GBP 1,500. All prices exclude VAT.
Book a free 30-minute scoping call
Tell us how many seats, which systems and where IT stands. You get a written recommendation on package and plan, and the questionnaire if you want to go ahead. Team Install from GBP 4,500 ex VAT, Business Install from GBP 7,500 ex VAT.