Do we build the integration everyone asks for, or the API that unlocks everything?

Every week someone asks for a specific integration. Zapier, Slack, Notion, HubSpot, whatever tool they happen to live in. And every time I say "not yet," I can feel a small amount of goodwill leaving the conversation.

Building the integration they want is satisfying and immediate. It solves a real problem for a real person, it's easy to announce, and it gives me a concrete thing to point at when someone asks what's new. But then next week someone else asks for a different integration, and I'm suddenly in the business of maintaining a growing library of connectors — each one fragile, each one a new thing that can break when they change their API.

The other path is to build one good public API and let the ecosystem do the work. Instead of me building ten integrations, I build the foundation and someone else builds the ten. But that requires documentation, stability promises, rate limits, auth, versioning, and a developer community I'd have to actually nurture. It's slower, less visible, and might result in nobody building anything.

I've watched companies drown in integration requests and slowly become a support desk for other people's APIs. I've also watched companies build a great API and never get the ecosystem they hoped for.

Which did you choose — the integration people ask for, or the API that lets them build it themselves? And did the ecosystem ever actually show up?