Post 3 — thesis-expanded

When we tell people we're starting a software studio in 2026, the next question is usually some version of "what are you building?" The honest answer takes longer than a sentence, and the short version sounds either too ambitious or too modest depending on which word lands first.

So here's the longer version.

The thesis has three parts.

One — AI is a leverage event, not a product category.

Most of what's been written about AI in the past three years frames it as a thing to add — a feature, a product line, a startup category. Add AI to your todo app. Build the AI-native version of X. Ship the wrapper around the model.

We don't think the interesting move is to build with AI in that sense. The interesting move is to use AI to build more, and smaller. Three people can now ship what used to require thirty. A team that would have struggled to maintain one product can credibly pursue four. The constraint that defined the last twenty years of software — "we'd love to fix that, but we don't have the headcount" — is loosening fast.

This isn't a market. It's a pry bar.

Two — The problems worth solving are the ones big companies won't touch.

There's a category of software problem that exists in every industry, looks the same everywhere, and reliably gets ignored by the companies large enough to fix it. The pattern:

• The market is too small to move a $10B revenue line
• The customers are too quiet to make noise (no one tweets about their tax-prep software being slightly less awful)
• The fix requires deep care about people who aren't decision-makers
• The reward for shipping it is "those users are slightly less miserable"

A startup with a 10-figure exit ambition cannot pursue these. The math doesn't work. A studio with a 20-year horizon and three products in flight can.

We pick from this category. Not because it's noble — though it sometimes is — but because it's the category where the leverage of small + careful + AI-augmented compounds best. You can't outscale Microsoft on a problem Microsoft cares about. You can routinely out-do Microsoft on a problem Microsoft has decided isn't worth caring about.

Three — The studio is the right shape for this.

A startup picks one bet. A studio picks several. The math is different.

Each Discerne product is a discrete bet. StoryPass is one bet (security UX). The next two are different bets. None of them is the studio's only shot. If StoryPass plateaus at 10,000 happy users instead of 10 million, that's fine — it didn't have to carry the whole house. The studio's success is the portfolio's success, not any single product's exit.

This shape changes how each product gets built. We don't have to chase metrics that would matter to a single-product company. We don't optimize for retention curves that justify a Series B. We optimize for the thing being good and the people using it being well-served. The economics of a studio with three modest successes look better than the economics of one moonshot that has to land.

It also changes the kind of work that gets done. A startup pours all its design energy into the one product. A studio can afford to spend a week getting the password generator's spacing right because the password generator isn't carrying the rent. The result is software that feels different — less aggressive, less anxious, less sales-y. Software that has the room to be careful because it doesn't have the burden of being everything.

What this looks like in practice.

Three products in the studio. One shipping (StoryPass). Two in development. A small team — three people. AI-augmented across the stack. Brazil-based. Founded 2025. Cash-flow positive in year one is the goal — not because we're allergic to VCs, but because the studio model only holds together if no single product needs to be a unicorn.

Writing about the work as we go. Open about what doesn't work. Patient about what we ship.

This is what we mean by "software on your side." Not as a tagline — as the operating principle that comes out of the three points above. Software made by people whose incentives are aligned with the people using it, because the studio model removes the incentive to be aligned with anyone else.

The full essay will go deeper on:

• What makes a problem "too small" for a big company, and how we identify them
• The role of AI specifically — what it lets the studio do that wasn't possible in 2015
• Why three products in flight is the right number, and why neither one nor ten works
• What we're explicitly not doing (no consulting, no agency work, no tools-for-other-startups plays)

Coming soon. Sign up at the bottom of discerne.co if you want to know when it's live.