Skip to content

Prototype

Multi Modal Agent Failure Handling

Built at the Google and Cerebral Valley hackathon. An agent orchestrating vision, text, and video models, with the actual objective being what happens when one of them fails.

Status
Prototype
Last updated
March 2026

01

So far

Orchestration demos well and degrades badly. Almost all the engineering went into timeout handling, partial results, and deciding when a missing modality means answer anyway versus stop. That part is never in the demo, and it is the only part that determines whether the system is usable.

02

Stack

  • LangChain
  • Node.js
  • Vision models

Contact

If you are working on something where being wrong matters, I would like to hear about it.

I am open to consulting engagements, research collaborations, and conversations that do not have a clear outcome yet.