Amazon AGI Lab Workshop: Visual Annotation + Verification Loop for Coding Agents
AI Engineer · youtube · 2026-10-12
AI Engineer published a workshop with Emile Baizel and Shruti Arora of Amazon AGI Lab on perception agents.
- Core idea: Two primitives form a loop — ① visual annotation: users select elements, draw, or drop pins on a page and export structured JSON feedback to a coding agent; ② visual verification: Amazon Nova Act checks the rendered site against design expectations and user flows, letting the agent inspect its own work instead of trusting its claims.
- Live build: Arora starts from a podcast landing page needing heading, color, and button changes, wiring a browser extension to a coding agent via a local bridge. Participants modify extension permissions, add an Apply Changes button, and turn exported feedback into an edit → verify → report workflow, with easy and medium difficulty paths. In the demo, an agent claims it added a duck icon and a browser verification pass actually checks the change.
- Extensions: A simulated Bee conversation as a design feedback source; audience ideas include spoken feature discussions, project planning, and recorded UI feedback.
- Resources: Full code at github.com/amazon-agi-labs/perception-agents-workshop.
More from coding & agent
- OpenJet claims Qwen3.8-27B coding runs 1.9x faster than llama.cpp on one RTX 3090 — JeremyCMorgan · 2026-10-23
- Building an autonomous SEO agent in Hermes with three skills: programmatic pages, Keywords Everywhere, DataForSEO — EXM7777 · 2026-10-23
- Building a sports betting AI agent with onchain execution and human approval by default: architecture and lessons — Dry-Comfortable-2678 · 2026-10-23
- Capy 0.5.0 lets Claude Pro/Max subscribers reuse their plan quota, adds OIDC cloud access — garrytan · 2026-10-23
- You.com CEO Richard Socher: legacy code is human slop, let AI clean it up — RichardSocher · 2026-10-23
- Let agents design a single-line glance component so you can track them from the sidebar — blixt · 2026-10-23