# Startup Spotlight: Confident AI turns production agent failures into tests for the next release > Source: > Published: 2026-10-26 16:13:20+00:00 # Startup Spotlight: Confident AI turns production agent failures into tests for the next release **Jeffrey Ip built DeepEval before Confident AI's hosted platform; Finom reports faster iteration, but its case study does not measure AI quality.** By [RuntimeWire Staff](https://runtimewire.com/author/runtimewire-staff) · Published Primary source: [Y Combinator](https://www.ycombinator.com/companies/confident-ai) ## Why it matters Confident AI is trying to turn open-source evaluation adoption into an enterprise workflow business. Finom says the platform let product managers run evaluations directly, reducing waits for engineering. [Confident AI](https://confident-ai.com/?ref=runtimewire) targets the work triggered when an AI agent fails in production: finding the cause and preventing the next release from repeating it. Co-founder and CEO [Jeffrey Ip](https://www.ycombinator.com/founders/1314014-jeffrey-ip?ref=runtimewire) ([@jeffr_yyy](https://x.com/jeffr_yyy?ref=runtimewire)) made that loop the product, connecting its open-source evaluation framework to a hosted system that can turn troublesome production traces into new tests.… When you purchase through links in our articles, [we may earn a small commission](https://runtimewire.com/affiliate-monetization-standards). This doesn’t affect our editorial independence.