Claude Code, set up on your machines and taught on your own work.
We install Claude Code and the tools beside it, connect them to the systems your team already runs, write the instructions and skills your work needs, and teach the team on real tasks: the week's posts, the ad copy, the blog, the SEO list, the outreach, the reports that run on their own.
This studio runs on it. Two people ship apps, stores and a client practice with a bench of AI agents, and the setup we sell is the one we use.
- This site
- 128 pull requests merged
May to October 2026, two people. Counted 19 October.
- The lead alert
- 469 lines, one Sunday
The site's own push notifications, web-push.ts and lead-alert.ts, written with Claude Code on 20 September 2026.
- The checkout
- Built, tested, live in three days
PayPal orders, capture and the webhook on this site, 17 to 19 October 2026.
- The bench
- Claude, Codex, Gemini
Each on the work it is best at, with a person deciding what ships.
How much time did you actually keep?
The first draft is only the start. Count the checking, corrections and interruptions too.
The work, not the demo.
The intake asks which of these you want. The setup and the training are built around the ones you pick.
Social media
The week's posts drafted from your own notes and captures, replies in your register, the calendar filled, and the scheduling done by a task that runs on Monday morning.
Google Ads
Search-term reviews summarised, negatives proposed, ad copy written against the landing page, and the monthly note drafted from the account export.
Blogs and long-form
Research pulled into an outline, the draft in your voice from your own material, an edit pass, the images briefed, the post filed where it goes.
SEO
Keyword research sorted into a page map, briefs written, the technical crawl turned into tickets a developer can take, Search Console read every month.
Outreach
Lead lists qualified, the first email written per person from what is public about them, the follow-ups queued, the CRM updated without anyone copying between tabs.
Support and the inbox
Replies drafted from your help pages and order data, the ones that need a person flagged, the rest ready to send.
Reports
The spreadsheet pulled, the numbers compared with last month, the note written, every Monday at eight, with a log of what it did.
Scheduled tasks
Any of the above on a timer, running on its own, with a log you can read and a person deciding what goes out.
The tools, in your accounts.
Every key stays yours. We connect the tools to what the team already runs; nothing new is signed up for unless the work needs it.
Claude Code- OpenAI Codex
Gemini- Google Workspace
- Microsoft 365
Shopify
WordPress
Google Ads- HubSpot
- Notion
- Slack
- GitHub
Product names and marks belong to their respective owners and are used to say what we set up, nothing more.
Three ways to start.
- $2,000one payment
Claude Code setup, one operator
Your machine set up and taught on your own work, in two sessions.
- Claude Code installed and connected to the tools you run, with every key in your own accounts.
- An instruction file and three skills written for your work: posts, ads, blog drafts, SEO pages, outreach, reports, whichever you do.
- One scheduled task running by the end of the second session.
- A written runbook and 30 days of email questions.
- Credited against the team workshop or the four-week program if the team follows within 90 days.
- $4,500to $7,500
Team workshop
Half a day or a full day, on your team's real tasks.
- Every seat set up and connected during the day, keys in your accounts.
- A shared instruction file and the first skills, written with the team.
- One real task per person finished with the tools before the day ends.
- A written playbook and 30 days of email questions.
Half a day for up to six people, $4,500. A full day for up to ten, $7,500. On Google Meet; on site at cost.
Open - $18,000one payment
Team enablement, four weeks
Everyone set up, two workshops, weekly office hours, the team's own skills and scheduled tasks built.
- Every seat set up and connected, week one.
- Two workshops on the team's real work, weeks one and three.
- Weekly office hours, and the skills and scheduled tasks built between them.
- A playbook, and 60 days of email support after the fourth week.
Up to ten people. $1,200 a seat beyond that.
Open
Afterwards, if the team wants the momentum to hold, the AI operator on retainer (from $5,000 a month), and the AI-agent integration ($7,500 to $20,000).
How it runs.
The intake
Five questions, two minutes. Who is on the team, what the tools should do, what you use today, which package, and when. We reply the same working day.
The setup
Claude Code and the tools beside it installed on each machine, connected to the systems you run, with every key in your own accounts. The instruction file and the first skills are written with you, for your work.
The work
Nobody watches a demo. Each person finishes one real task with the tools before the session ends, and the team leaves with a shared library it built.
The playbook and the support
What was set up and why, in writing, with the runbook for each skill and task. Then 30 or 60 days of email questions, and weekly office hours inside the four-week program.
Fair questions.
Do we need our own Claude subscription?
Yes. The seats are yours on Anthropic's own plans and the keys stay in your accounts; we do not resell seats or sit between you and the tool. The intake asks what you already pay for so the setup fits it.
Is this for developers?
No. Marketers, founders and operators get the most out of it, because their work is the kind the tools finish well: posts, ads, blogs, SEO lists, outreach, reports. Developers are welcome; the setup is the same and the skills are different.
Which tools do you set up?
Claude Code first, because the studio runs on it and we teach from experience. OpenAI Codex and Gemini beside it where they fit a task better. Cursor or Copilot if the team writes code.
What happens to our data?
Everything runs in your accounts on your machines. We keep nothing after the sessions except the playbook we wrote for you, and that is yours too. Each tool's data policy is linked in the runbook so the team can read it.
Remote or on site?
Google Meet by default, which keeps the price where it is. On site anywhere at cost, for the full-day workshop and the four-week program.
Can we start with one person?
Yes. The one-operator setup is $2,000, paid on the spot, and it is credited against the workshop or the program if the team follows within 90 days.
Read what useful AI training should leave you able to do, or compare ChatGPT, Claude and Gemini for a small business.
Tell us how you would use it.
Five short questions so the first call is about your work and not about ours. We reply the same working day, Monday to Friday, with the first slot and a short list of what to have ready.
Prefer to talk first? Book a 30-minute fit call. Thirty minutes on Google Meet, Monday to Friday, to find out whether we are the right shop for the job. No deck, no pitch.