Your office, anywhere · one to two days on site · up to twelve people · two engineers at the tables
Your people already have prototypes: a GPT that drafts the proposal, a spreadsheet macro, a workflow someone pasted keys into. Each one feels productive and none of them has a database, a login, a review or an owner. In your room the team learns the difference by building the other thing: one central database the company owns, reached only from the server, with an admin behind a login and every change reviewed.
They arrive with a stack set up and leave with a full stack application on the company data, a workflow running through it and a managed agent on a schedule.
Laptops stay closed for the first hour. One prompt per block after that, engineers at the tables, and the system on your domain before the room closes.
An interview with Claude writes the plan for your system on your data. Before lunch: the repository, the project, the database, a login, a live form and one admin page. The afternoon builds the whole system from the plan, in your brand, on a branch, merged through a pull request.
Ninety days of realistic data and a dashboard on top. A QA plan and a senior code review. Then the first managed agent: it runs on a schedule, reads the company database, acts within its permissions and logs every run. Every owner opens the live system on their own phone before close.
Not a prototype in a chat window. A deployed application with a login, an admin and your real records, on your own domain, before the room closes.
Keys in environment variables, the database reached only from the server, every change through a reviewed pull request. The thing security asks for, built in from day one.
An agent that runs on a schedule, reads the company database, acts within its permissions and logs every run. The kind a business can build on.
Product manager, developer, QA and devops, configured on your repository, so the team ships the next feature on Monday without us.
One call to scope the problem, a written quote within two days, then a date.
Corporate retreats run inside companies, on their data. The recaps belong to them.
Ask us for references and we will put you in touch with a founder or GM who has run one.
It is quoted per engagement after a thirty minute call: the number of days on site, the size of the team and whether one or two engineers are needed set the price. You receive the quote in writing within two days. Travel for the engineers is itemised, nothing else is.
In your own Supabase project, on your own GitHub organisation, deployed to your own Vercel account. Keys stay in environment variables, the database is reached only from the server and the admin is behind a login. We work inside accounts your company owns and we hold nothing when we leave.
The person who owns the problem, the operator who runs the process today, and whoever will keep building afterwards. Up to twelve people at the tables. Nobody needs to be technical; engineers handle the integrations and the infrastructure.
Each builder brings a laptop with Claude, GitHub, Vercel and Supabase accounts created and the stack set up from our one page guide. We check it on a call the week before. The company also picks the one problem the system will solve, so day one starts with planning, not a debate.
The system is live and the four agents are on the repository, so your team ships the next feature on Monday. If you want senior engineers on tap for the first thirty days, human tokens cover that, metered like Claude tokens, no retainer.
Once at no charge with thirty days notice. Inside thirty days we rebook within six months and only pass on engineer travel that cannot be refunded.