Most founder folklore starts at launch. The harder stage is earlier: figuring out what is worth the next decade. In isolation it feels like a squiggle. Talent makes it worse, because every capability is one more path that must be ruled out.

The old advice was to raise and throw things at the wall. That is a weaker bet now. Software has already eaten the easy fruit. AI makes it cheaper to build and easier for thin traction to become a local maximum. A low-conviction fundraise can consume years.

Patterns noticed at W26 Demo Day of 199 Startups:

Key signals from the batch:

Sequoia's Julien Bek: the next trillion-dollar company won't sell the tool — it will sell the work itself.

For every $1 spent on software, companies spend $6+ on services. AI just turned that massive budget into a startup goldmine.