How One RCA Stopped a Costly Scale-Out: A Talk with Two CTOs
CTO Navdip Bhachech and NeuBird CTO Vinod Jayaraman on the RCA that stopped a costly database scale-out, RCA fatigue and agent memory. Watch the FLOCK '26 talk.
At FLOCK '26, NeuBird co-founder and CTO Vinod Jayaraman sat down with Navdip Bhachech, CTO at DemandTec. Navdip first brought NeuBird, the AI SRE, into Bedrock Analytics, and has now brought it to a second company.
At Bedrock, the team was convinced a database slowdown was a traffic problem and was ready to scale out the database. NeuBird traced it to a concurrency bug in the database connection pool, down to the line of code. Scaling out would have cost more and only delayed the problem, which is why the story shows what a single actionable root cause is worth.
From there they discuss RCA fatigue, keeping operations in step with code written by coding agents, and why agent memory has to be scoped by system, environment and customer so agents do not recall the wrong thing. Navdip's request for every agent: be brief, be brilliant, be gone. His wish for NeuBird: investigations that understand whether a system is single-tenant or multi-tenant.
In this conversation
Filling the DevOps and ops skills gap
Traffic, or a concurrency bug? The RCA that changed course
How one RCA avoided a costly database scale-out
RCA fatigue
AI-written code is outpacing operations
Scoping agent memory, and why memory is not knowledge
Related Resources
See NeuBird in action
Root cause in minutes, not war rooms.
