← All projects
Concept for a0 · by Anirudh Devella

Catch the bad generation before the user sees it

a0 turns a prompt into a mobile app. When the agent gets stuck fixing the same error, the user watches it loop, then leaves. This shows where an eval layer would sit, and replays one session with and without it.

Fictional example. The people, numbers and events are made up to show the method. This is an outside concept, not a0's product or data, and not affiliated with a0.

Built for a0 · the short version

When the agent loops on the same error, the user watches it fail, then leaves.

In the replay, the eval gate catches the repeat error on the second try and the user ships an app at 3:10. Today the same session ends in a closed tab.

The data is made up. To make it real, I'd need generation logs with build results and errors, plus each user's sign-up channel.

Who built it

Anirudh Devella, data scientist. Before this, in production:

  • 8.3% forecast error at Bluelight (YC)
  • $14M+ in invoices benchmarked at Bluelight
  • AI Coach for 10,000+ athletes at Brandlete
  • 150K leads scored at CoverRight

“I’d hire him again.” Alex Dixon, Co-founder, Bluelight (YC), managed Anirudh directly. Read it on LinkedIn

01 · Where it fits

A gate between the agent's output and the user's screen

Today whatever the agent builds goes straight to the preview, broken or not. The eval gate checks each result first. If it fails, the agent retries with a different approach, out of sight. The user only sees a working build or a plain explanation.

02 · Same prompt, two sessions

Replay: a first-time builder asks for a habit tracker

Left is what happens today when the agent hits a package error it can't shake. Right is the same session with the eval gate. Press play.

or step with

Today

0:00

With the eval gate

0:00
03 · What the gate checks

Four checks, cheapest first

Most failures are caught by fast, plain rules. A model is only used for the last check. Every verdict is logged with the user's sign-up channel, which is also how you find out which marketing brings builders who actually ship.

Next step

My first week with your real data

  1. Measure how often sessions hit the same error twice, and what share of those users never return.
  2. Ship checks 1 and 2 (build passes, same error twice) as a hidden retry.
  3. Report ship rate by sign-up channel so marketing spend follows builders who ship.