Prism AI: Agentic observability — agents watch production logs and customer session videos then file issues automatically
Jun 11, 2025
Key Points
- Prism AI's founders manually operated their product during Y Combinator batch, watching customer videos and filing issues themselves while marketing it as AI-powered before building the actual agent-based system.
- The startup deploys AI agents to monitor production logs and customer session recordings, automatically filing issues in Linear or GitHub and sending Slack reports to developers.
- Prism targets individual developers and early-stage startups through self-serve, currently reaching 23 users mostly from the YC community, with founder Rajith having previously worked at code review company Greyline.
Summary
Prism AI deploys AI agents to monitor production systems for developers. The agents read logs, watch video recordings of customers using software, and automatically file issues in Linear or GitHub while sending reports to Slack.
The company targets individual developers and early-stage startups rather than enterprise buyers. The product is self-served and has 23 users, drawn primarily from the YC community.
Founder Rajith disclosed an early operational detail: during the YC batch, the first version ran manually. The founders themselves watched customer session videos and filed issues and Slack messages to users while marketing it as AI-powered. The actual agent-based product came later.
The founding team brings relevant experience. Rajith previously worked at Greyline, a code review company. One co-founder was a software engineer at Palantir and is raising capital while closing a funding round. The third co-founder worked in product at Johnson & Johnson. All three met at Georgia Tech. One founder left a compensation package worth $850,000 annually at Palantir to join the startup.