Why do LLM apps break in production — and how can you stop them before users notice?

In this episode, Nir Gazit (co-founder of Traceloop, ex-Google/Fiverr, YC W23) breaks down how his team is solving AI’s biggest failure mode: observability. You’ll hear how Traceloop became the go-to black box recorder for LLMs, powering companies like Cisco, IBM, Miro, and more.

We cover:

• Real-world LLM horror stories
• How OpenLLMetry hit 2M+ downloads
• Miro’s strategy for safely deploying GPT-4.1
• What AI teams can learn from traditional dev observability
• The roadmap to self-improving agents

If you're building with LLMs or planning a GenAI rollout, this is essential listening.

Where to find me:
• David: Linkedin
• Email: [email protected]

Want to stay ahead of AI's impact on jobs and future opportunities? Join 18000+ tech leaders and founders who are shaping the future of AI here:

• Newsletter: Join Here