A letter on the first twelve.

I started Buildable because I kept watching the wrong companies get called moats and the right ones get called toys.

For most of the last decade, the question for a SaaS investor or operator was: is this product good? The question worked because building software was expensive enough that the answer usually told you everything else — distribution, retention, even defensibility. If you could ship it, and people paid, the moat was implicit in the work.

That question stopped working sometime in 2024. AI made the surface of most SaaS products legible — the components, the flows, the data model, the integrations — to anyone willing to look. A landing page is no longer a brochure. It is a specification. And a specification, in 2026, is most of a rebuild.

So the question I actually wanted answered was different: what survives when the surface is reproduced? Not whether the product is good. Whether the company would still be a company if a competent team rebuilt the visible parts in a weekend.

That is the question Buildable tries to answer, one product at a time, with three coordinated scores: how rebuildable the surface is, what survives the rebuild, and how cleanly the rebuild lands in Lovable's architecture specifically. Three numbers, one verdict, every report versioned and disclosed.

The first twelve.

Wave 1 is twelve companies — Notion, Okta, Datadog, HubSpot, Dropbox, Zendesk, DocuSign, Asana, Salesforce, Atlassian, GitLab, Klaviyo. They were not picked because they are the most important SaaS companies in the world, though several are. They were picked because each one tests a specific claim I wanted to make about where moats actually live now.

The corpus is not a sample. It is an argument. The argument is that buildability and moat are independent axes — that you can have a product whose surface is almost trivially reproducible and whose company is nearly untouchable, and that the inverse is also true, and that the interesting work is in the diagonal.

Twelve was enough to make the argument visible. Fifty is what we need to make it empirical. The methodology is currently at v1.1.0, with calibration weights for Lovability Fit drawn from operator elicitation rather than logged outcomes. v1.2.0 unlocks the day the fiftieth build is logged. The progress bar on the about page is honest about where we are.

What I learned writing them.

Three things, none of which I expected.

First, the products that look easiest to rebuild are almost never the ones easiest to replace. Notion's surface is a recognizable block editor. Its moat is a decade of accumulated user content and the fact that nobody ever exports it. Dropbox is a sync folder. Its moat is being on every machine your team already owns. The surface is the bait. The moat is somewhere else entirely, and the analysis has to look for it on purpose.

Second, the translation cost into a specific architecture matters more than I thought. A high Buildability Index is not a permission slip. Some products decompose cleanly into Lovable's primitives — others require so much custom infrastructure that the rebuild is closer to a fork than a translation. Lovability Fit is the score I most want to be wrong about, and the one I am most careful to label as independent estimate, not certification.

Third, the language you use to publish this kind of analysis is the entire product. Call it a clone and you have picked a fight. Call it reverse-engineering and you have invited a lawyer. Call it structural analysis with versioned methodology and an independence disclaimer, and you have a discipline.

What I am asking for.

Two things, depending on who you are.

If you build software for a living and the framing here is useful to you — read a report, paste a URL, disagree with a score in public. The methodology improves through pressure, not applause. Every score is comparable only within the same major version. v1 is on the record. v2 will be different, and the diff will be argued.

If you work at Lovable — I would like to talk. Buildable is independent of you by design (the disclaimers are real, not ornamental), but the Lovability Fit dimension is a measurement of your architecture, and measurements get better when the thing being measured pushes back. I am the Lovable Ambassador in Brazil. The work is already in your direction. I would rather calibrate it with you than around you.

Either way, the next twelve are already being audited.