I had Claude, Gemini, Antropic, and a couple of other providers wired into Creative AI Writer.
Felt clever. Lots of options for users, hedge your bets across providers, pick the best model for each task. The works.
In practice: I was spending all my time managing API differences, normalising outputs, keeping up with model deprecations, and debugging edge cases that only showed up with one provider. None of that is the product. It’s just overhead.
Just pick one
went with OpenAI. Done.
It’s not that the others are bad — they’re not. But OpenAI is a solid all-rounder. The models are capable, the API is stable, the tooling ecosystem around it is mature. For a writing app, that’s more than good enough.
The creative tasks I care about — story prompts, scene generation, character beats — GPT handles all of it well.
What I got back
Dropping the other providers immediately gave me back time I didn’t realise I was losing. Fewer conditionals, fewer test cases, fewer things to monitor.
The app got simpler. The codebase got smaller.
Should have done it earlier.
the SDKs that handle multiple providers help, but they still need to cater for edge cases. I’m not sure you can really have one interface to rule them all…
Links:
- Creative AI Writer: https://creativeaiwriter.com
- OpenAI API: https://platform.openai.com