Hi to 6 in 3
Hi, thanks for reading! I quit Microsoft AI over 3 weeks ago to build what I wish existed - and test 6 new ways to democratize AI every two weeks with real users.
Why I left
In mid-April, I was offered a major project in Microsoft AI. Instead of taking it, I chose to leave Microsoft after nearly 2 years— even after being encouraged to consider a sabbatical instead— to build full-time.
I’d gone into product after reading The Sprint Book & loving the idea of quickly testing hypotheses. At Microsoft, I built & improved metrics for Copilot in Edge & Windows, saved then released a 7x retention consumer product, and worked across notifications, gaming, experimentation, ML-driven notifications & more, balancing 3+ projects at a given time & starting a cross-org team to encourage more user research, as the first PM in my org to interview users directly in 2 years.
Over the past two years I’ve explored computer use agents, RLHF modalities, accessibility, and organized one the world’s first AI Agents for Nonprofits hackathons, and I didn’t see a team running fast cycles to test new ideas in neglected spaces or in the spirit of democratizing AI research, so I thought I’d “start” my own.
What I’ll be doing
I’ll be running six sprints in three months, each one designed to test, fast, if the ideas can deliver real value — in speed, savings, or accuracy — for real users. The goal isn't just to explore: it’s to earn trust, find traction, and converge toward something scalable or impactful as new tech’s unlocked many more ways to help.
I’m not betting on a single feature. I’m betting that velocity compounds. Even if none of these ideas scale, I’ll have helped real people, built real things, and learned a lot about a process to find what works and what doesn’t quickly, starting with GV’s design sprint process. I designed an incubator program I wished existed, with pre-mortems, kill switches, anti-goals (re: Sahil Bloom), weekly retros, and lava cakes after every sprint.
Why follow along
There’s a lot of noise right now that you could be reading online so my commitment to you is I’ll share what I learn, be concise, and you’ll come out of each sprint’s retro as if you were a fly on the wall of a mini-MBA. Most people quit with funding or to join another company- I’m doing neither, instead betting on myself. If you want to follow along, I’m publishing the insights, (many) failures, and structured hypothesis tests from each sprint here, taking shots on net with the goal of decreasing time to validate hypotheses & find need. Feel free to leave any feedback!
My incubator is inspired by, among others:
The Sprint Book (https://www.thesprintbook.com/the-design-sprint)
Creating your own MBA (https://dianaberlin.com/posts/the-mba-decision-guide-for-product-managers)
Nikil Viswanathan’s ETL talk on how he tried new ideas quickly (youtube.com/watch?v=4cmDFNXT1Fs)
Best, Zach