Teach
We install the way. Your team builds on it.
Nothing ships from us. No automated tasks, no AI assistants, no integrations. What ships is the way to build with AI: how to spot which work to automate first, how to grow it into something bigger, the review process that catches problems before they hit production, and the routines your team will run after we leave.
Front-loaded intensive. Four to six weeks, mostly on-site (or all on-Zoom if that's your default). Then we leave. Your team builds on top of what we installed; what they built keeps running because they own it.
Section I · What gets installed
The way. The routines. The quality bar. Not the code.
i · the way
How to build with AI
Taught as a working idea your team applies, not a slide deck. They walk away knowing how to spot which work to automate, when to grow a small automation into a real AI assistant, and what "ready" looks like at each step.
ii · routines
Data, testing, quality, adoption
Regular routines for the four basics. What to capture, when to review, how to spot when the AI starts being wrong, when to ship, when to pull. The routines run on a calendar your team owns.
iii · quality bar
The shipping checklist
A one-page checklist that defines "good enough to ship". Your experts grade each other's work against it. It's the spine of quality control after we leave.
iv · runbook
The starter runbook
A living document, built with your team, in your repo, in your voice. Describes how to add a new automation, wire up an AI assistant, and run the routines. The team updates it after we leave.
Section II · In practice
A SaaS product team. Six-week intensive.
The product team had a senior engineer who wanted to lead an internal AI program but had never set one up before. We ran a six-week intensive: the way week, routines week, quality-bar week, two paired-build weeks, a retrospective week. We were never the ones doing the build.
6 weeks intensive · 22 automations + 4 AI assistants shipped after we left · 0 PRs reviewed by us
End of quarter: 22 automations and 4 AI assistants in production, all owned by your experts, running through their own quality reviews. They shipped six more in the four weeks after we left. Foundry left. The system kept running.
Section III · When to choose Teach
You have the build. You have the engineers. You don't have the way.
- Your team can write automation already. What's missing is the way to actually run a program: which one to build first, when to grow it, how to keep quality up. You need the approach, not the code.
- You want to set this up without us in the building forever. A defined-end engagement that leaves you running it.
- Lowest cost per automation long-term, because everything from week seven onward is yours — built by your team, reviewed by your experts.
Some clients run a parallel Build on a critical workflow during the intensive, or follow up with Ride shotgun for the first quarter the practice is theirs.
Section IV · Talk to us
If your team needs the framework more than the artifacts, send us the shape. We'll scope the intensive.
foundrysolutionsai@proton.me →