Why we haven't spent a dollar on CreatorOS yet

Early on we set a simple, non-negotiable rule for building CreatorOS: zero dollars get spent on the product — domains, paid API tiers, developer program fees, upgraded plans, anything with a real cost — until there's a first real sale. Once there's revenue, it gets reinvested. Until then, everything has to be built on free tiers or not built at all.

That rule has real, visible costs of its own. For a while it meant our transactional email had to go out from a shared, restricted sending address that could only reach our own inbox — meaning no real user could receive a password reset or a product update email. It meant Google's OAuth verification for the sensitive YouTube scopes we need was stuck, because Google requires a real homepage URL as part of verification, and we didn't have a domain to point at. It meant Sign in with Apple was off the table entirely, since Apple's developer program is a flat $99/year regardless of whether anyone's paying us yet.

What the rule didn't block

Everything that costs nothing but time kept moving. We rebuilt authentication around Google and Discord OAuth (Discord's developer portal is free), added a `plan` field to the data model so a future paid tier has somewhere to land even though nothing reads it yet, and wired up audience syncing to our email provider's free tier so the infrastructure would be ready the moment sending was unblocked. None of that needed a credit card.

We did eventually make one exception: buying the creatoros.onl domain, a small, deliberate spend that unblocked both the email-sending restriction and the OAuth verification requirement at once. That wasn't the zero-spend rule quietly eroding — it was a conscious, one-off call, made explicitly as an exception rather than a policy change. The rule itself still stands for everything else.

The honest reason for a rule like this isn't discipline for its own sake. It's that spending money on a product with no revenue is a bet, and we'd rather every early bet be made in engineering time — which is recoverable if we're wrong — than in cash, which isn't.

Want to see what we’ve actually built?