“Earning within two weeks of launch, and now generating so many leads that Dougie has hired someone just to keep up with them.”
Buying five Claude licences takes a minute.Getting five people to use them is the job.
For businesses standardising on Claude who need their team actually using it. Summone runs the workshop, maps your documents and processes, sets the governance, and builds the Claude Team architecture your people work in. Construction, professional services and manufacturing, from a Glasgow base.
What the rollout is worth.
Your numbers, not ours: put in the quote or budget you are actually weighing up. Based on a 46-week working year. Most teams we train report 10+ hours saved per person per week; the default here is half that on purpose.
Licensing is not adoption, and the gap between them is where the money goes.
Most Claude rollouts stall in the same place. The seats get bought, two people use it properly, the other three go back to doing it the old way, and six months later somebody asks what the subscription is for.
That is not a tooling problem. It is a problem of nobody having mapped what the business actually does, decided which of it Claude should touch, and written down who signs off the output. That mapping is the work, and it is what this engagement is.
This page is deliberately detailed. Every question below has been asked by a real managing director before signing, usually over three or four weeks of email. Read it before the call and we can skip straight to your business.
Four steps, in this order, every time.
What Claude Team actually includes.
A lot of what is published about Claude Team seat types is wrong, including on sites that ought to know better. The most common error is that Claude Code and Cowork are a Premium feature. They are not. Standard seats include both. Premium buys you more usage, not more features.
Figures below are Anthropic list prices in US dollars, excluding tax, checked against Anthropic's own help centre on 31 July 2026. UK pricing varies with exchange rate and VAT treatment, so treat these as the shape of the cost rather than your invoice.
| Standard seat | Includes Claude, Claude Code and Cowork. $25 per seat per month, or $20 billed annually. |
|---|---|
| Premium seat | Same features, higher usage allowance. $125 per seat per month, or $100 billed annually. |
| Seat counts | Minimum 2 seats, maximum 150. Above that you are into Enterprise. |
| Usage limits | Per member, not pooled. Standard is roughly 1.25 times Pro per five-hour session, Premium roughly 6.25 times. |
| Mixing seats | You can mix Standard and Premium in one organisation, and buy usage credits past the included limit. |
| Connectors | Microsoft 365, Google Drive, Gmail, Calendar, Slack and GitHub, with enterprise search across them. |
Licences and consulting are two different bills.
These get conflated constantly and it kills otherwise sensible decisions. The licences are paid to Anthropic. The consulting is paid to Summone. They are not the same money and they do not do the same job.
The licences are the cheap part. What decides whether this works is whether five people are still using it in November.
| Licences to Anthropic | Five Standard seats comes to roughly £80 to £100 a month, depending on rate and billing term. |
|---|---|
| Discovery workshop and onboarding | From £3,950. Includes the workshop, the process mapping, the prioritised roadmap and the initial Claude Team build. The deliverable is included, not extra. |
| Monthly retainer | Ongoing training, build work and review sessions. Priced against the scope agreed at the end of the workshop. |
| Minimum term | Three months on the retainer, then 30 days' notice at any point. |
| In scope | Prompt and template work, knowledge base upkeep, new workflow builds, monthly review, and support for your team by email. |
| Out of scope | Anthropic licence costs, third-party software, and bespoke application development, which is quoted separately. |
The section you should read twice.
This is the part most buyers care about most, and rightly so. Putting tender pricing, client correspondence and staff records into an AI system is a real decision with real exposure, and it should not be waved through on a sales call.
Nothing commercially sensitive moves until the position below is confirmed with you in writing. An NDA is signed first, as standard, before we look at a single document.
| Model training | On the enterprise tier, your data is not used to train models. This is contractual, not a setting somebody might toggle. |
|---|---|
| Where data sits | Confirmed in writing before anything sensitive moves, along with retention settings and how to change them. |
| UK GDPR and DPA | Position documented for your records, so your own compliance file is complete rather than resting on our word. |
| Access control | SSO, Domain Capture and just-in-time provisioning, with role-based permissions so estimating cannot read HR. |
| Spend control | Limits set at organisation level and per user, so usage cannot quietly run away from you. |
| Leavers | Seats and access revoked centrally. Knowledge stays in your organisation, not in an individual's personal account. |
Nothing AI-generated becomes an official document without a named human signing it.
That is especially non-negotiable for Health and Safety material. Claude drafts. A competent person reviews, amends and signs. The controlled, approved version lives in your systems with your version control on it, not in the AI.
In practice that means RAMS packs and method statements are drafted against your existing approved templates and then reviewed by whoever currently reviews them. Tender specifications, labour and quantity calculations, and HES and NTS handover records go through the same route. The approval chain does not change. What changes is how long the first draft takes.
We write this into the governance document at the workshop, name the sign-off holder for each document type, and make it visible in the tool itself so nobody has to remember the rule.
You own it, and the point is that you need us less.
Everything built for you belongs to you. Prompt libraries, templates, knowledge bases, workflows and any bespoke application built as part of the engagement. If you stop working with Summone tomorrow, all of it stays with you and keeps working.
The retainer exists to make your team more capable every month, not more dependent. Every session is training as well as delivery, and the intention is that a year in your people are doing most of this themselves and calling us for the harder problems. A consultant who builds themselves into a permanent dependency has sold you a liability.
Baseline on day one, reviewed every month after.
The baseline is captured in the workshop before anything is built, because a return you cannot show against a starting number is just a feeling. The measures are yours, not a generic dashboard.
| Typical measures | Hours per tender, tenders processed per month, time to produce a RAMS pack, time from enquiry to quote out. |
|---|---|
| Workshop done means | Roadmap signed off, and every person on a seat using Claude on real work rather than test prompts. |
| Monthly review | Numbers revisited at each session, with the roadmap reordered if the value is somewhere other than we expected. |
Yes, we are a Claude consultancy.
Worth stating plainly, because search engines currently describe Summone as Microsoft Copilot focused and that is a fair reading of what has been published rather than of what gets built. The Copilot material is client-facing work written for Microsoft-centric organisations. It is not the delivery environment.
The delivery environment is Anthropic, end to end. Claude Code is used daily and every live client application is built and maintained in it. This page you are reading was built that way.
| Experience | 26 years in enterprise and solutions architecture, much of it at board level. |
|---|---|
| Enterprise clients | Includes NatWest and Virgin Money. |
| Day-to-day tooling | Claude Code, Claude Team and Cowork, in production, not in a pilot. |
| Insurance | £1m professional indemnity, held since the bank work. Certificate on request. |
| Who you get | Steven, throughout. No juniors, no handoffs, no account manager between you and the person doing the work. |
Training developers instead? Claude Code training → · AI consulting → · Enterprise AI consulting → · AI training →
Judge us on what we ship.
“We find your buyers, write the emails, and send the outreach. Enquiries land in your inbox. You do nothing.”
“I had an idea and I just wanted to bring it to life. Summone supported me through the whole build, design, planning and testing. It was not just my idea, it was their support.”
Asked and answered.
Does Claude Team include Claude Code and Cowork?
Yes. Standard seats include Claude, Claude Code and Cowork. This is worth stating because several third-party guides claim Claude Code is a Premium-only feature, and it is not. Premium is a usage upgrade rather than a feature upgrade. Verified against Anthropic's help centre on 31 July 2026.
How much does Claude Team cost per user?
Standard seats are $25 per seat per month, or $20 per seat if billed annually. Premium seats are $125 per month, or $100 billed annually. These are Anthropic list prices in US dollars excluding tax, checked on 31 July 2026. UK pricing varies with exchange rate and VAT treatment. For a five-person team on Standard seats, budget roughly £80 to £100 a month.
What is the difference between Standard and Premium seats?
Usage, not features. Both include the same tools. Standard gives roughly 1.25 times the Pro allowance per five-hour session, Premium roughly 6.25 times. Limits are per member rather than pooled across the organisation. You can mix both seat types in one organisation and buy usage credits past the included limit, so the sensible approach is to start everyone on Standard and upgrade only the people who hit the ceiling.
How many seats do I need?
The minimum is 2 and the maximum is 150, above which you are into Enterprise. In practice the right number is the people who will use it weekly on real work. Buying seats for people who might use it one day is the fastest way to make the subscription look like waste at renewal.
Is our data used to train Claude?
On the enterprise tier, no. That is contractual rather than a setting somebody could switch on by accident. Before anything commercially sensitive moves, we confirm in writing where the data sits, what the retention settings are, and the UK GDPR and Data Protection Act position, so your own compliance file stands up on its own. An NDA is signed before we look at any of your documents.
Who owns the prompts and workflows you build for us?
You do, all of it. Prompt libraries, templates, knowledge bases, workflows and any bespoke application built during the engagement belong to your business. If you stop working with Summone, everything stays with you and keeps working. The retainer is there to make your team more capable each month, not more dependent.
How do we stop AI-generated documents being treated as approved?
By writing the rule down and building it into the tooling. Nothing AI-generated becomes an official document without named human sign-off, and that is non-negotiable for Health and Safety material. Claude drafts, a competent person reviews and signs, and the controlled approved version lives in your systems under your version control rather than in the AI. We name the sign-off holder for each document type during the workshop.
How long does a Claude rollout take?
The discovery workshop is a day. The initial build and onboarding typically runs over the following few weeks, depending on how much of your process is already documented. After that it is a monthly rhythm: build, train, review. Three-month minimum on the retainer, then 30 days' notice at any point.
Do you work with construction and trades businesses?
Yes, and the document types are the point. Tender specifications, RAMS packs, method statements, labour and quantity calculations, programme work, HES and NTS handover records. These have real approval chains and real liability attached, which is exactly why the sign-off rules matter more here than in a marketing team.
Are you a Microsoft Copilot consultant or a Claude consultant?
A Claude consultant who has also delivered Copilot work for Microsoft-centric clients. The Copilot material on this site is client-facing work written for organisations already standardised on Microsoft 365. It is not what Summone builds in. The delivery environment is Anthropic end to end: Claude Code daily, and every live client application built and maintained in it.
Let’s talk about your business.
A straight conversation with Steven. No pitch, no juniors, no handoffs. He will map where you are losing time, what it is worth, and what to do about it, before you spend a penny.
steven@summone.co.uk
Glasgow, working with UK businesses remotely and on-site