A post saying founders who hand-coded their products for three years now feel an order of magnitude behind teams that started this year has passed 320,000 views on X, and it confuses fast code with fast learning.

Josh Elman wrote it after talking to several of those founders. The gap he describes is real: an agent can write a feature in an afternoon, and a team typing by hand cannot match that.

But the three-year-old company holds things a team founded this year does not. It knows which edge cases customers actually hit, why the odd decisions in the architecture were made, and which users already pay. None of that lives in the code. A full rewrite by agents produces new code quickly and starts the learning again from zero, with paying users finding the old bugs a second time.

Elman himself calls the move a risk and a reboot, and that is the better frame. Keep the product, hand the coding to agents, and spend the first weeks writing the context into the repo: the tests, the reasons behind past decisions, the list of edge cases the team already learned. An agent with that context works like a senior engineer who has been there three years.

So the older teams hold the one thing agents cannot generate on their own, and the job now is to put it where the agents can read it.