Skip to content

Day 5 · MVP presentations + close

Friday 29 May 2026 · Tokyo · Senior audience + workshop participants

The closing day. The morning is the engineers’ moment — each one presents the MVP they built across the week. The afternoon is a discussion on AI guardrails and a reflection on the week.

Morning · ~2-hour block · Engineers present their MVPs

Each engineer presents the MVP they built during the workshop. The format is short and focused — the goal is to see twelve different working systems, each tuned to the engineer’s own interest.

  • Setup — one shared screen, engineers cycle through
  • Per MVP — ~5 minutes: what it does, who it helps, live demo, the agent pattern behind it
  • After each demo — a few minutes of Q&A from the senior audience

The variety is the point. Twelve engineers, twelve different MVPs — incident collectors, guardrail rules, custom front-ends, specialist agents, small offices. All built on the same Claude Agent SDK and orchestration patterns walked across Days 1 – 4.

A short closing section after the MVPs:

  • The week’s arc — what was built, what was opened, what changed
  • Brief look at multi-company offices on the same substrate
  • Open Q&A with the senior audience

Architecture overview

Afternoon · ~2-hour block + close

A discussion, not a presentation. The team listens.

  • How an organisation should think about AI safety in its own deployments
  • Which guardrail patterns from the week transfer to existing domains
  • How to govern agent-generated code as fleets scale
  • Reflections from each participant, then close

Guardrail Lab reference

What success looks like

  • Every engineer presents a working MVP they built themselves
  • The senior audience sees twelve concrete examples of what an agent team can do
  • One or two MVPs become candidates for follow-up — extended, productionised, or shared internally
  • The week has produced material the team will reference for months

Materials

Next: Workshop overview — The full week at a glance.