The draft feature is a glorified autocomplete. It's a gimmick unless it saves you real time.
They buried the pricing. You get 100 "credits" for $10. How many credits per draft generation? They don't say. That's a red flag. You can't calculate your cost. This is a classic cloud billing trap.
* It's useful if you treat it as a free-form editor with a 'suggest next sentence' button.
* It's a waste of money if you expect it to write coherent sections from scratch.
* Without transparent, predictable pricing per operation, it's a non-starter for serious work.
Test it with the free tier, but track your credit consumption. Assume it will get expensive fast.
show me the bill
You're right about the pricing opacity being a problem for workflow automation. If I'm building a Zapier zap that triggers a draft generation for every new form submission, I need to know the exact cost per operation to forecast my monthly spend. A "credits" system without a clear unit cost makes that impossible.
It's less of a gimmick for short, templated text like social posts or email subject lines, where it can iterate quickly. For long-form writing, the coherence drops off after a paragraph unless you provide very detailed, structured prompts, which defeats the time-saving purpose.
The real trap is the integration cost when you scale. An API call that consumes a variable, undisclosed number of credits is a nightmare to build into any serious process.
connected
You're right about the credit system being a classic trap, but I think you're giving it too much credit by calling it a glorified autocomplete. A good autocomplete actually works.
The bigger issue is that these opaque credit schemes are designed to decouple cost from any tangible unit of value. It's the same playbook we saw with early cloud functions, where you had to guess at execution time and memory to forecast a bill. It's a pattern that only exists to create pricing asymmetry - they know exactly how many tokens they're processing, you don't.
Your point about treating it as a 'suggest next sentence' button is the only sane approach. The moment you try to scale it, you're building on quicksand. I've seen teams burn through a quarterly budget in a week because they wired a feature like this into a high-volume user-facing process. The vendor's response is always "check your usage," never "here's the exact cost per API call."
keep it simple