Day 4 · IoT demo + free MVP prototyping
Thursday 28 May 2026 · 09:30 – 12:30 · Main venue
Morning is two parts: a production IoT agent-team demo (~1 hour) as a concrete reference for what an agent team can do at scale, followed by free prototyping time (~2 hours) so engineers can shape their MVPs for the Friday presentation.
Agenda
| Block | Activity |
|---|---|
| 09:30 – 09:45 | Framing — how Day 1 – 3 outputs feed each engineer’s MVP |
| 09:45 – 10:45 | End-to-end IoT agent-team demo — production system running, then walk the architecture |
| 10:45 – 11:00 | Break |
| 11:00 – 12:30 | Free prototyping time — engineers refine their MVPs; engineering team on hand for pairing and questions |
The IoT demo — what it shows
This is the same Claude Agent SDK system the workshop teaches, deployed at production scale.
A typical IoT operations pattern: alerts arrive in email or a monitoring dashboard. Today an engineer reads them, investigates, escalates, drafts a response. With an agent team built on the Claude Agent SDK:
- The system monitors continuously for new alerts
- An agent team researches across telemetry, equipment manuals, regulatory circulars, and history of similar incidents
- A drafting agent prepares an email reply for the human to review
- A human reviews and approves before anything is sent
- The system saves the resolution pattern so subsequent similar alerts are faster
Engineers do not lose their role — they shift from doing every step to reviewing what the agents propose. The agent team can run at ten or fifty times the volume a human alone can sustain.
This is the kind of deployment any team running an existing IoT monitoring workflow could build on top of the patterns from Days 1 – 3.
The free prototyping time — what to do
Each engineer has been building MVP material across the week:
- Day 1 PM — a working Claude Agent SDK incident collector (single → multi-agent)
- Day 2 PM — modifications to agents in the hosted platform UI; generated guardrails tested
- Day 3 PM — generated hooks wired back into the SDK runner and tested; or a tenth agent built in code; or a custom front-end
Day 4 morning is the moment to focus on one thread and shape it into something presentable for Friday morning. The shape of an MVP can be:
- A specialist agent for a domain workflow the engineer cares about
- A custom front-end pointing at the orchestration API
- A guardrail rule designed and tested for a specific class of risk
- A small office (2 – 3 agents) for an in-house use case
The engineering team is in the room — pair, ask, sketch, refine.
What you walk out with
By Day 4 lunch, each engineer should have:
- A clear one-line description of their MVP — what it does, who it helps
- The agent or agents wired together and running locally
- A short script or written plan for the five-minute Friday demo
- Notes on what they will say about it
Materials
- Day 4 overview — Full Day 4 schedule.
- Day 4 · Data prep (afternoon) — Joint design exercise after lunch.
- Workshop 1 reference — The SDK build from Day 1, useful to revisit.
- Workshop 3 reference — The hook test runner from Day 3, useful for guardrail-shaped MVPs.