Talk: Your AI Agents Are Already Broken In Production (You Just Don’t Know It Yet)

[imshashank]

First name

Shashank

Last name

Agarwal

Startup domain

https://noveum.ai

GitHub repo or gist to share (optional)

No response

Twitter / X handle (optional)

itsshashank

Talk summary

AI agents are already silently failing in production — most enterprises just don’t realize it yet. At Noveum.ai, we’re building autonomous reliability infrastructure for AI agents: replay testing, continuous evals, synthetic voice testing, regression detection, and PR-based auto-remediation workflows. The hardest problem has been closing the loop between detecting failures and safely validating fixes before deployment. I’ll share real-world lessons from deploying these systems across enterprise voice agents, customer support automation, and multi-step AI workflows.

[github-actions]

@imshashank 🎉 Your talk is approved and live on the homepage!

Please RSVP to lock in your spot: https://luma.com/event/manage/evt-XgCxpVXEsGHIVHD

RSVP-ing Yes shares your Google account email with the organizer (that's how we'll send you logistics — slot, venue, speaker chat, etc.). No email collected anywhere else.