From Incidents to AI Products: What Operations Taught Me About PM
From Incidents to AI Products: What Operations Taught Me About PM
I didn't start in product. I started in the operations trenches, as a Product Support Specialist and then a Major Incident Manager, chairing outage bridges, querying databases at odd hours, and writing the RCA after everything had already gone wrong. People assume that's a strange path into product management. I think it's one of the best.
Running incidents teaches you where systems actually break
When you've managed the conference bridge during a live outage, you stop trusting the happy path. You learn that real systems fail in the seams, the timeout nobody handled, the integration that silently drops, the spike no one watched for. That instinct never left me, and it's exactly what you want shaping a product roadmap.
In e-commerce and travel, the unhappy paths are a large share of real usage. A booking that half-completes, a payment that errors, a search that returns nothing useful, these aren't edge cases, they're where trust is won or lost. Years of incident management trained me to design for them first.
Communication under pressure is half the job
During major incidents I owned the communication to internal and external stakeholders while the fire was still burning. You learn to say what's true, what's unknown, and what happens next, clearly, calmly, and fast. That's not an ops skill. That's the core of product leadership.
A roadmap is just communication over a longer timeline. The same discipline that keeps an outage bridge from descending into chaos keeps a cross-functional team aligned on why we're building something and what we'll do if it doesn't work.
Bringing AI into the picture
I went back to study Microsoft Azure OpenAI and AI because the next frontier of the work I already love, search, conversion, personalisation, runs on AI-driven insight. The grounding mattered: understanding what these systems can and can't do keeps me from treating AI as a magic word in a roadmap.
The goal isn't to bolt AI onto a product for the optics. It's to use it where it genuinely sharpens a decision, better intent matching, smarter personalisation, faster diagnosis of where a funnel leaks, and to stay just as suspicious of its failure modes as I learned to be of any system in production.
What the whole path adds up to
Support taught me how products actually behave once they leave the demo. Incident management taught me to communicate clearly when it matters most. Product taught me to turn all of that into decisions people can ship against. AI is the newest lens, not the whole picture.
The path wasn't a pivot, it was an accumulation. Every role added something I still use, and the job now is to keep all of them in focus while building products that hold up on a bad day, not just a good one.
Chabi skipped presentations and built real AI products.
Chabi Anand was part of the April 2026 cohort at Curious PM, alongside 18 other talented participants.