"Recede in te ipse quantum potes; cum his versare qui te meliorem acturi sunt."
— Seneca, Epistulae Morales, VII — Withdraw into yourself as much as you can; keep company with those who will improve you.
In 2025, the AI industry announced a new model roughly every eleven days. New tools, new benchmarks, new paradigm shifts, new frameworks that would make everything before them obsolete. If you read every announcement, evaluated every release, and tested every tool, you would have done nothing else.
I know this because I tried it for about three months.
The problem with following AI news as a founder is not that the information is bad — some of it is genuinely significant. The problem is that consuming it feels like building. It produces the sensation of progress while actively preventing it. You spend four hours reading about a new model's capabilities and go to bed feeling informed, having shipped exactly nothing.
"Dum differtur vita transcurrit."
— Seneca, Epistulae Morales, I — While we delay, life passes.
Seneca's instruction is precise: withdraw into yourself. Not passively — actively. Recede is a verb of deliberate motion. Go back to the interior. He is describing a discipline of selective attention, not retreat from the world but retreat from the noise that lives on the surface of it.
For the AI founder, this means something concrete: maintain a short, fixed list of tools. Build deep familiarity with each one. Do not switch when something new appears unless you have a specific, evaluated reason. Do not follow every announcement. Treat your toolstack as a commitment, not a buffet.
My current infrastructure is three tools I have used consistently for more than a year. I know their failure modes. I know how to write specifications that get reliable outputs from each one. I know where they break and how to work around it. This knowledge is not available at the start — it accumulates over time, through repeated use, through observing what works and what doesn't.
The founder who switches models every month never accumulates this knowledge. They are always in the evaluation phase, never in the mastery phase. They are everywhere. Nowhere.
Seneca's second letter to Lucilius continues the theme of his first: stop scattering yourself. He talks specifically about books — the danger of reading too many authors without letting any of them become part of you. "Distringit libros multorum lectio," he writes. The reading of many books is a distraction. Pick one thing. Go deep. Let it become yours.
This is the Founder on AI posture. FOA is not about using the most advanced tools — it is about using fewer tools with greater precision. The competitive advantage is not tool selection. It is operational depth: the accumulated understanding of how to get specific, repeatable outputs from a small set of reliable systems.
The world will continue to announce new models. The benchmarks will improve. There will always be something newer, something apparently better, something worth at least a look. Seneca saw the equivalent of this in the intellectual culture of Rome — a market of ideas, schools of philosophy, visiting rhetoricians, new texts arriving from Alexandria. The pressure to stay current was real then too.
His answer: choose your sources with care, then go deep. Ignore the rest, not from incuriosity but from discipline. Recede in te ipse. The work that matters is done in the interior, where the noise does not reach.
The AI founder who builds durable systems is not the one who evaluated every model. They are the one who chose three tools and built mastery with them while everyone else was watching announcements.
Withdraw. The news will keep without you.