Workshops · Engineering

Your fastest path to
AI-native engineering.

A hands-on workshop that turns your engineers into agent builders — and leaves a real harness running on your own stack. Built around Claude Code, MCP, and the company brain we deploy.

SEE THE PLATFORM →
How It Works

Bring a real build. Leave with it running.

A focused workshop scoped to a real workflow — not a class project. You keep the harness, the runbook, and a benchmark.

∫01
Select your team
The engineers — plus the PM who owns the workflow — who will make agents part of how you ship.
∫02
Bring a real build
No toy projects. Pick a workflow worth automating — a flaky regression suite, a migration, a support triage pipeline — and we scope it before we start.
∫03
Ship it live
By the end, that workflow runs on a harness wired to your systems, with evals and observability — not a slide deck.
Ways to Engage

Three ways to run it.

Every format ends the same way: a working harness on your stack.

ENGINEERING TEAM
Team Workshop
A focused multi-day workshop for one engineering team.
  • →Onboard to Claude Code + MCP
  • →Stand up the universal MCP layer
  • →Ship one harness against a real workflow
  • →Remote or onsite — all IP stays yours
BOOK A CALL
MOST POPULAR · CROSS-FUNCTIONAL
Cross-functional Squad
Everything in Team Workshop, plus the people who run it after we leave.
  • →Everything in Team Workshop, plus:
  • →The PM or ops owner of the workflow joins
  • →They learn to operate and extend the harness
  • →Collaboration patterns that outlast the workshop
BOOK A CALL
CUSTOM
Custom / Embedded
We embed with your team, scoped to your stack and roadmap.
  • →We come to you, on your timezone
  • →Curriculum scoped to your stack and AI roadmap
  • →Flexible team size and schedule
BOOK A CALL
The Curriculum

What your team will master.

Grounded in Claude Code, MCP, and the agent patterns we run in production.

THE CURRICULUM
∫01AI-native coding with Claude Code
The agentic loop (Explore → Plan → Code → Commit), context management, custom commands, and hooks.
∫02MCP & connectors
Wire Claude to your systems — GitHub, Jira, Postgres, internal APIs — and the universal MCP layer every harness shares.
∫03Agents, subagents & orchestration
Delegate to isolated subagents, run multi-step plans, and build the harness pattern: reason → tool → reason.
∫04Evals, observability & safety
Eval rubrics, tracing, and guardrails so the system earns trust before production.
∫05Build sprint
Ship the real workflow onto a harness wired to your stack.
∫06Handoff & scale
Runbook, ownership, and the path from one harness to many on one synced brain.
Who Runs It

Operators, not trainers.

The same people who deploy and run our own harnesses.

∫01
Practitioner-led
They have shipped production agent systems — they know the failure modes, the deploy gotchas, and the evals that matter.
∫02
Pattern recognition
When an agent loop will not stop or an eval will not calibrate, they have seen it before and point you at the fix.
∫03
Hands-on in the sprint
On the floor for spec reviews, eval calibration, and the production push.
Book a Workshop

Book your workshop slot.

A focused engagement: we wire your AI clients into one shared company brain, live. Tell us where you'd start and we'll send times — we take on a limited number of new workshops each month.

NOW BOOKING · LIMITED SLOTS30-minute scoping call · sized proposal in 24h
Booked and run on the same platform we operate our own company on.