Make money doing the work you believe in
Arize AI raised $131 million and ships features the same day a customer reports the bug.
Their CPO walked through the actual workflow. A PM spots a high-priority issue in the morning. They prototype a fix in Claude Code by lunch. An engineer reviews. Production by end of day. The sprint cycle that used to gate every feature request now runs in hours.
She proved it live. Built a working PM agent during our conversation that pulls GitHub issues, scores them by priority, generates a daily triage report, and runs on a cron loop. Then instrumented it with tracing in a single command and had Claude write the first eval.
The whole build took about an hour. From "I have a product idea" to "I have a self-improving agent running my daily triage."
Code got cheap enough that the bottleneck moved to product taste. The PMs who can identify the right problem at 9am and prototype a solution by noon are operating at a velocity that makes the old sprint-based workflow look like filing paperwork.
At AI native companies right now, PMs who can do this are indistinguishable from engineers. That's the bar moving in real time.
Here's how to get there:
Ship your first PR: news.aakashg.com/p/pm-g…
Get the scaffolding: news.aakashg.com/p/pm-os
Learn from Arize: youtube.com/watch?v=DL-…
Model OpenAI: open.spotify.com/episod…
Of course, there are limits. At some places, this isn't the best use of a PM. Don't ever blindly copy another company.
Choose the right model for you.
