I got tired of finding gaps
after I opened Figma.

So I built a Figma plugin that reads a PM brief before design starts and tells you exactly what’s missing — edge cases, undefined states, unclear success criteria — with a completeness score and the specific questions to ask.

Briefs that looked complete until they weren’t.

At AIR iQ, I work inside a B2B travel booking platform that moves fast. A brief lands on Monday. Design is expected Wednesday. Handoff specs by Thursday. In that window, nobody’s formal job is to ask: “wait, what happens when a user’s session times out mid-booking?”

The gaps only surfaced when I was three screens deep and couldn’t finish a flow because a key state hadn’t been decided. Empty states. Error conditions. Success criteria that were really just vibes. Every time I had to stop and go back to the PM, a little bit of design time — and trust — got lost.

The third time I had to ping mid-design.

The PM didn’t know either. They had to loop in three people, schedule a call, and two days later I had an answer that changed the flow I’d already designed. I wasn’t frustrated at the PM — I was frustrated that we had no process to catch this before it happened.

The fix needed to live inside Figma. That’s where designers are when they first read a brief, and that’s where the gap hits them. A separate tool wouldn’t get used. It needed to be right there, instant, before the first frame.

Fast model, thin backend, Figma native.

The plugin UI is built on the Figma Plugin API — clean, minimal, intentionally out of the way. The backend runs on Cloudflare Workers so API keys stay server-side, cold starts are near-zero, and there’s nothing to maintain. For the AI, I used Groq running Llama 3.3 70B — fast enough that the response feels immediate, and it handles structured output reliably, which matters when parsing a brief for specific gap categories.

A completeness score and the questions that matter.

You paste the brief. BriefCheck reads it across three lenses: edge cases the brief doesn’t address, UI states that haven’t been defined, and whether there’s a clear success criteria to design toward.

What comes back isn’t vague notes — it’s a score and a specific list of questions to bring to the PM before Figma opens. That conversation happens when it should: before the design, not halfway through it.

I launched it on LinkedIn with an Apple-style silent demo video. It got more traction than anything I’d posted before — not because it’s a clever tool, but because every designer immediately recognised the problem.

The version of me from six months ago would have opened Figma and hoped for the best. Building this changed how I work — and made me realise that the most useful design tools are the ones that make the right conversation happen at the right time.