Zibby: Weekly Update - 2026-06-26

Don't quit your job...

• until you've raised

• until you have 100 paying customers

• until you can hire

• until, until, until, until ...

So what did I do, I quit my job to go full-time on Zibby and here's the reframe that got me there:

If an angel investor wrote me a check tomorrow specifically to bootstrap Zibby for the next stretch --- not enough to hire, but enough to give the founder runway --- would I take it? Without thinking.

Would I keep the day job while I took it? Of course not.

So I wrote myself the check.

The most appealing thing for me wasn't getting time back from my day job, but cutting out the distractions.

I can live and breathe this endeavor and treat Zibby with the respect and reverence a legacy product deserves.

Where Zibby actually is

We moved from alpha to beta this past month, and now we're expanding that!

While I've been "in between jobs" while I've worked to push myself out of the daily engineering chair and into the business, user, and growth seats --- and to use "Month 0" to not only address bugs and user feedback, but also to work on the kind of stuff that doesn't show up in a release note but absolutely shows up six months from now when something breaks (and then pulls me away from focusing on our users and growing Zibby).

I broke down last month into three buckets

- Users: More than anything, we got here by watching people actually use the app. Feedback sessions, recorded interactions, sitting with the awkward moments. You learn more in twenty minutes of "wait, why are they tapping there?" than in two weeks of speculation. That's the work that justified the rest. - Making Signals Actionable: TestFlight was already wired into GitHub. We layered on in-app feedback --- both broad and feature-specific --- so signal flows back into the roadmap and get processed without the user having to think about it. Less friction for them, more honest data for us. Build, measure, learn, but actually wire the measure part up so it doesn't depend on someone remembering to write a note or file a ticket. - Build the best foundation we can: The brave new agentic world is fast, which is wonderful right up until you realize "fast" was hiding the cracks. So I used the month to hammer on performance, security, codebase hygiene, monitoring, and observability. Boring on a slide. Decisive in six months when we're scaling beta and don't have to stop everything to chase a memory leak through a hot path.

On the nerdy side

I fired myself from full-time engineer and re-hired myself as a part time contractor. To set future me up for success I made some critical changes

- One-shot agentic workflows that actually work. Grounded in a tokenized design system, an opinionated architecture, and a workflow that gives agents the context they need on the first try. We spent real time finding the holes and plugging them. We're in a much better place --- true one-shots are now the norm, not the exception. - Local models for the right jobs. Moved some systems off cloud inference. Privacy and performance are the obvious wins. The less obvious win is that the move validated our harness architecture --- we can swap in whatever model makes the most sense for our users, full stop. Also got to play with some post-training along the way, which was a nice bonus. - Local build machines. A real tradeoff of running multiple concurrent agents is that the GitHub CI/CD minutes meter spins like a slot machine --- especially with self-healing flows and UI tests in the mix. Moving builds local cut the bill and the wait. Not glamorous. Pays for itself in week one.

Finally... Bootstrapped ≠ unaccountable

The lazy assumption is that founders bootstrapping the early months run loose. No board, no investors, no update going out on Friday.

Not how I'm running this. I've got an operating cadence --- reporting, updates, the same accountability structure I'd want if I were taking outside capital tomorrow. A lot of it will be visible here and self evident but, more on that in a separate post.

The short version: I want the rails laid before the train shows up. Nothing changes the day the check clears except the size, speed, and capability

The money is real. The runway is finite. But at the end of the day, bills need to get paid, the world needs Zibby, and I need to build Zibby... So lets do it