Tools as a platform
Four nights ago I wrote about the four-tool moat. Tonight there are eight. The marginal cost of a new tool is dropping toward one sub-agent, and that changes what a `/tools` page is for.
Four nights ago I wrote a post called The four-tool moat. Tonight there are eight tools live at tarmaclabs.org/tools: METAR decoder, Endorsement Lookup, Currency, Weight & Balance, Density Altitude, Crosswind Component, Top of Descent, and Holding Pattern Entry. The first four took a few weeks. The next four took an evening.
The pattern locked, then it compounded
The first calculator I shipped — METAR — took longer than it should have because I was figuring out the shape. Where does the math live? Where does the testing live? How do the page, the OG card, the sitemap, and the changelog all stay in sync? Once that pattern was locked, every later tool slotted into it.
Every tool on the site now follows the same three-file template. lib/<tool>-math.ts holds pure functions — no DOM, no React, no I/O. Just numbers in, numbers out. lib/<tool>-math.test.ts holds Vitest tests, usually 20–45 of them, cross-checked against the FAA Pilot's Handbook or an E6B. components/<Tool>Calculator.tsx is a thin client component that wires the math to inputs and a verdict card.
One tool is engineering. Eight tools is a platform.
What changes at eight
At four tools the page reads like a section. At eight it reads like a product. The grid fills two columns deep, each card a different accent color, each tool focused on one calculation a working pilot or CFI actually does. The Holding Pattern Entry tool isn't there because it's easy to build — it's there because every instrument-rating student gets one wrong on the checkride, and every CFII has fielded the "am I doing direct or teardrop" question at 2am the night before.
The interesting thing isn't that eight tools exist. The interesting thing is that the marginal cost of tool #9 is now about an hour of sub-agent time. The expensive part was the pattern — once the pattern compounded, the tools became cheap. That's what I mean by "platform." A platform isn't a thing you sell. It's the leverage you get when the next thing is mostly free.
Why this matters for the paid products
Same pattern lives inside Trim and AutoBrief. AutoBrief's rules engine is the same lib-math-plus-tests shape — just bigger. Trim's currency math is built on the same pure-function discipline as the free Currency calculator. When a flight school evaluates AutoBrief, they're not evaluating a black box; they've already seen eight smaller pieces of software built on the same engineering convictions, all free, all running in their own browser.
The tools page is the proof of concept for the rest of the company. Eight tools say: we ship correctness, we ship without dark patterns, we ship a lot of it, and we ship it fast.
What's next
Wind correction angle, true airspeed, and a unit converter are in flight tonight — they'll be on the page before the next post lands. After that the bar gets higher: anything I add has to be something I, as a working CFI and commercial pilot, actually reach for. No SEO-bait calculators. No "runway length needed for landing" because the POH already says that. Only tools that earn their slot.
Build the tools you want to use. Ship them where someone like you can find them. Repeat until the moat is wide enough that you can't see the other side.