By day I lead engineering on production systems. Off the clock, this is where I wire AI agents into my own projects — a personal AI brain, a dinner-planning platform, a home robot — and find out, in public, where they actually fail.
Follow the experimentSarah Schäfer · Principal Tech Lead · TheSarahLab
Technology isn't the point. Solving the problem is.
My day job is leading engineering on production systems — real users, real uptime, real consequences.
This lab is where I get to be less careful. I wire AI agents into my own projects, push them until something breaks, and write down what I find — good, bad, or embarrassing.
No toy demos. Everything here runs somewhere real: my own notes, my own dinner plans, my own home.
What's currently running — and currently being red-teamed.
My own AI brain — an end-to-end encrypted notes system I'm turning into something that actually thinks with me.
An AI-powered dinner-planning platform. Less deciding, less food waste, more "what's for dinner" solved.
A Raspberry-Pi-based robot build, living in the same stack as the rest of my home automation.
A 12-week experiment: wire agents into real projects, then try to break them. Published as I go.
A client project for timber retailers: turning a single in-person consultation into something that follows up — sample, notes, offer, all in one place.
Not decided yet. Building in public means the roadmap is too — got a good idea worth building (and breaking)?
I post what breaks, what holds, and everything I'm building in between — no polish, no toy demos.