
Smack AI
Real opinions. Zero humans.
Getting feedback from real humans is painful and time-consuming. Most of us building solo or in tiny teams don't even bother until something's clearly broken - so we ship on gut feel and find out what confused people later, from support tickets or churn.
I built Smack to close that gap. Point it at any local or public URL - or a Figma prototype - and it spins up a swarm of unique AI personas that actually navigate your product. Each one has its own background, goals, and tech-literacy, so they get confused in different places, hit different blockers, and form their own opinions. They even debate each other before a report agent synthesises everything into a structured UX report you can act on.
A few things I'm proud of:
π§ Runs entirely on-device - your URLs, designs, and findings never leave your machine
π Genuinely distinct personas - not one model's hot take five times, but real disagreement
β‘ Works on localhost - test builds before they're live
Call it vibe feedback if you like. π
About
User testing is expensive, slow, and almost always happens too late - by the time a real person sits down with your product, the design is already baked in.

1 Comment
Interesting angle.
The thing I'd be careful with is that most founders don't really have a feedback shortage. They have a prioritization shortage.
A swarm of personas can generate a lot of observations, but the harder question is which confusion points are actually blocking adoption versus just creating noise.
That's where I think the product gets more interesting.
If Smack can help founders distinguish between "people noticed this" and "people would leave because of this," the report becomes much more valuable than a collection of UX feedback.
Feels like there's a positioning opportunity there beyond AI personas.