This is a synthesis of public discussion on Figma’s own community forum, not a tested or maker-announced fact, and not a survey of every Figma user. Read enough of the forum’s Share Your Feedback and Report a Problem sections from the first half of 2026 and one complaint keeps resurfacing in thread after thread, months apart, written by people who don’t appear to know about each other’s posts: Figma Make’s AI credits get consumed fastest not by generating something new, but by trying to fix what the AI generated wrong the first time.
What designers are saying
The clearest example is a single thread started in March that grew to well over a hundred replies and thousands of views, unusually high engagement for the forum. The original complaint is specific: most AI usage isn’t discretionary experimentation, it’s correcting bad or incomplete output, and the credit system charges for that correction exactly as if it were fresh work. Repliers across that thread, from brand-new accounts to people on full enterprise seats, describe the same shape of problem: burning through thousands of monthly credits in hours rather than weeks, with one person describing roughly 70 percent of their prompts going toward fixing what the AI got wrong rather than building anything new. No reply in that thread defends the credit system as reasonable, the disagreement in the room is entirely about how much more headroom would be enough, not whether headroom is needed.
A separate, smaller thread makes the same underlying point from a different angle: a user reports the AI failing roughly half the time, returning what they describe as fake or irrelevant interfaces while still reporting success, and Figma’s own support response in that thread attributes the inconsistency to prompt quality rather than disputing the failure rate itself.
A third, more recent thread shifts the complaint from cost to visibility. When Figma rolled out a persistent AI agent panel in the left sidebar, several people independently objected to it displacing the Assets icon they were used to clicking, calling it disruptive to existing muscle memory, and asked how to turn it off. Figma’s own support reply in that thread confirms individual users currently can’t hide it, only enterprise admins can disable AI features org-wide, which left individual complaints in that thread unresolved.
Why it matters for designers
If you’re evaluating Figma Make for real production work rather than a one-off experiment, the pattern here is a reasonable thing to budget around before you commit to it: treat the credit allowance you’re quoted as covering iteration and correction, not just first drafts, because that appears to be where most of it actually goes in practice. It’s also a useful frame for judging AI-assisted design tools generally, a tool that looks affordable based on “how many things can I generate” is a different proposition once you account for how many of those generations you’ll need to immediately redo.
What to watch out for
This reflects people who were frustrated enough to post on Figma’s own feedback forum, which self-selects for complaints over quiet satisfaction, so it isn’t a representative survey of every Figma Make user, plenty of people using it for lighter tasks may never hit these limits. It’s also specific to Figma Make’s credit-metered AI features, not a verdict on Figma as a whole or on AI design tools generally. And because this is Figma’s own forum rather than an independent community space, Figma support does sometimes reply directly in these threads, so treat the acknowledgments quoted above as customer-support responses to individual reports, not as the company’s considered position on the pattern as a whole.
