Annual: 2025

Annuals are the cathedral of the blog. Once a year, you stop, you take stock, you write the long thing. This is the first one I've written for myself, in public, with my name on it. It will not be the last. The shape of a year is most legible after it's over, but the shape of a life of years is only legible if you keep writing them down.

What follows is not a report card. It's a working document — the things I built, the things I learned, the things I'd do over. I'm writing it for the version of me sitting down to do this exercise next December, who will need to remember what mattered.

I. What I built

The biggest thing I built in 2025 was a rhythm. I shipped product on Mondays and Thursdays, wrote on Wednesdays and Saturdays, and took Sundays seriously. The rhythm produced — across the year — fourteen substantive product releases at Trayd, forty-one essays here and on Sage Ideas, and a body of operational knowledge that I'd previously only suspected I had.

The next biggest thing I built was a team that doesn't need me. We hired three engineers and one designer in 2025. None of them have asked me to make a technical decision in the last sixty days. That's not because I'm distant — it's because the system around them now resolves most decisions before they ask. The seatbelt rule. Reversible deploys. Document-then-decide. The infrastructure of low-trust collaboration that produces high-trust output.

The infrastructure of low-trust collaboration that produces high-trust output is the only infrastructure worth investing in.

The smallest thing I built — and the one I think mattered most — was a daily review habit. Three questions, every evening, in a cheap notebook: what did I do, what did I avoid, what did I learn. Two minutes. The compounding has been ridiculous.

II. What I learned

I'll list these as aphorisms with receipts. The aphorism is the lesson; the receipt is the story that taught it.

1. Reversibility beats correctness. Receipt: the migration story that opens "Why we refuse to ship anything that can't be rolled back in 30 seconds." Designing for undo turned out to be a deeper unlock than designing for getting it right the first time.

2. You learn more from publishing badly than from drafting well. Receipt: the three essays this year that I almost killed in editing went on to be the three with the most replies. The version I wanted to delete was the version other people needed.

3. The team you can hire is downstream of the system you've built. Receipt: the senior engineer we hired in Q3 told us she joined because she read our deploy docs. The docs were the recruiting funnel; we just hadn't noticed.

4. Calendar protection is the highest-leverage investment. Receipt: I tracked deep-work hours weekly. The weeks I had four blocks of three hours produced more than the weeks I had twenty meetings. This isn't a surprise. It is, however, one of those things that you have to keep re-learning.

III. What I'd do over

I would have killed two side projects six months earlier than I did. I knew the AI photo organizer wasn't going to win in May. I let it limp until November because I'd already announced it. The cost of admitting you're wrong is always less than the cost of staying wrong on schedule.

I would have published the Trayd arc earlier. I waited until the company was "ready." The company is never going to be ready. The arc was always going to start when I started writing it.

I would have hired the second designer. We hired one and pretended she could do the work of two. She delivered, and quit, and we deserved that.

IV. The next year

Three concrete bets I'm making for 2026, public enough that you can hold me to them:

The first bet is on focus over breadth. I'm going to publish fewer things, longer, with more time inside them. Twelve essays a year, twelve dispatches, one annual. The compounding from depth is undervalued.

The second bet is on infrastructure that ships itself. Every important decision at Trayd in 2026 needs to leave a paper trail that future hires can read without me being in the room. We're going to write the playbooks before we need them.

The third bet is on the people. I'm going to spend more time on the team we have and less time looking for the team we don't. The hires we made in 2025 have not yet had their best year here. I want to be the reason they do.

The cathedral takes a year to build. Then you tear it down and start the next one. I'll see you in the new year.