AI x Token Engineering

Why this session

AI agents can already support real parts of the token engineering lifecycle — research synthesis, spec drafting, simulation scaffolding, mechanism stress-testing, monitoring live parameters — but practice is scattered across individual teams, mostly undocumented, and untested against each other.

This session treats that gap as the topic itself, rather than one person presenting a finished answer nobody has yet.

Runs as Day 2 of the three-session arc — after the Breakout kicks off on Day 1, before everything closes out together on the TERSE day (Day 3), where this session also gets a dedicated readout slot.

Format

1. Practice Shares

   Day 2 of 3 · ≈20 min

   A few practitioners each give a short lightning share of their own process: what they use AI agents for today, what works, what doesn't, and where they've been burned by trusting an agent's output too far.

2. Collaborative Mapping

   ≈20–25 min

   Together, the room sketches a shared Token Engineering Process Machine — idea → research → spec → prototyping → modeling → implementation → go-live — and marks, at each stage, what can realistically be agent-assisted and where a human-in-the-loop checkpoint is still needed.

3. Wrap-up & Documentation

   ≈10 min

   Consolidate what the room converged on, and where it openly disagreed, into a living, shareable artifact rather than letting the discussion evaporate.

How to get involved

• Have a process, workflow, or hard-won lesson to share? Propose a short practice-share slot below — that's the only way to actively contribute beyond attending.
• Otherwise, no sign-up needed: everyone who shows up takes part in the collaborative mapping, there's no group to pick. RSVP on Luma just so we can plan for numbers.

Propose a topic RSVP on Luma — link TBD

Proposed topics

No topics published yet — check back once the first is approved, or propose one.

Recap

EthCC[10] recap will be added here after the event — check back after April 11–15, 2027.