◆
No second decision
The arrangement that suits five people still suits five hundred. You do not re-platform during the growth that caused the problem.
Early companies make an infrastructure decision twice: once when nothing matters, and again at the worst possible moment when everything does. The second one is the expensive one, and it is avoidable.
Used from first prototype through to the scale where people ask who runs your infrastructure.
What this looks like in practice
The arrangement that suits five people still suits five hundred. You do not re-platform during the growth that caused the problem.
Generous enough to carry a real product, not a tutorial. You start paying when you have something worth paying for.
No platform team required. The work is building the product, which is the only work you can afford.
What you build remains portable. That is worth more to a founder than any discount.
Growth-stage engineering time is almost entirely consumed by things that were fine at the previous size. Infrastructure should not be one of them.
Ask any engineer at a company between fifty and two hundred people what they are working on, and a surprising amount of it is replacing something chosen when the company was ten people. The cheapest infrastructure decision is the one you make once.
Every branch a real environment, which changes how a small team reviews work.
The marketing site and the product on the same footing, shipped by the same people.
Internal tools a founder can build in an afternoon and still rely on later.
An arrangement that stands up to technical due diligence rather than complicating it.
We are forty people now and have never once had an infrastructure migration on the roadmap. I did not realise that was unusual until I spoke to peers.
Yes — it is sized to carry a real product rather than a demonstration. You start paying when the product has the traction that justifies it.
Nothing structural. The arrangement that suited the prototype is the one that carries the scale, which is the entire argument for choosing it early.
No, and most teams here do not have anyone. If you reach the size where a platform team makes sense, it will be because you chose to, not because you had to.
Then you leave. What you build stays portable and exportable, and we would rather earn the renewal than rely on the cost of departure.
Move in stages with the old arrangement still running, and cut over only once the replacement has carried real traffic.
Get capacity when the experiment is ready, keep what produced a result, and put models in front of users without a separate platform.
Keep everything at full fidelity, query it without a project, and stop deciding today what you will want to ask next year.
Ship the application your users touch, on an address that stays put, with every version one click behind you.
Build the tools your own teams need, at a cost that makes small improvements worth doing.
Give your engineers a platform that removes decisions rather than adding them, without a platform team to run it.
Begin with one workload. Keep what you have running. Decide whether to continue once you have seen it carry something real.
No migration required to begin. Nothing to tear down if you stop.