Moda
Monitors AI agent production traces and turns the mistakes it finds into regression evals.
Moda reads every production trace from an AI agent and flags failures that don't throw errors, like fabricated order numbers or a confirmed refund with no matching invoice, grouping them into problems ranked by how many users they hit. Custom signals written in plain language catch specific patterns, such as an agent giving a UTC time to a customer in a different timezone, with Slack, Linear or email alerts when the hit rate crosses a threshold. Before shipping a prompt or model change, you can replay it against the exact traces that failed and compare pass rate, latency and token cost.
What holds up
- +Groups recurring failures into ranked problems with linked traces, instead of a flat error log
- +Replay compares a prompt/model change against traces that previously failed, with pass rate, latency and token deltas
- +Custom signals are written in plain language, not a query DSL, to catch team-specific failure patterns
Mind the limits
- −No pricing is shown on the page — it's demo-request only
- −Built specifically for agent production traces, so it adds little for teams without agents already in production
Picked here? Add the badge
Put it on your site. It links back to this page. Free for every listed product.
<a href="https://stillworks.dev/products/p/moda/"><img src="https://stillworks.dev/badge/moda.svg" alt="Picked by StillWorks" width="230" height="54"></a>