Free AI Headshots
Team
1 PM, 1 Engineer, 1 Designer, 1 ML Engineer
Tools
Figma, Jitter, Adobe
Platform
iOS, Android, Web
Year
2022-2025
My Role
Sole Product Designer
The third product I designedbuilt alongside the main product, not after it. One tool, one job: let anyone try AI headshots without signing up, paying, or committing to anything. The free generator wasn't an afterthought. It was a deliberate entry point we designed from the start, knowing the category had a trust problem. One photo in, one watermarked headshot out. No login. No credit card. No friction at all. The design problem wasn't technical. It was calibration. Make the experience feel real enough to create desire, but scoped enough that it leads somewhere. The entire page had one goal: turn a skeptic into someone who just saw their own face in a professional headshot and wanted more.
Scope of Work
Starting Point
Why This Existed
The Design Challenge
Most free tools make one of two mistakes. They give too little a blurry preview, a heavily degraded output and the user leaves underwhelmed. Or they give too much, a full experience, no reason to upgrade and the user leaves satisfied.
Neither converts.
The real design job here wasn't layout or interaction. It was calibration. Finding the exact point where the output is impressive enough to create want, but incomplete enough to create action.
That line is harder to draw than it sounds.
What We Built
One headshot. Four crop shapes. No login.
Upload a photo, get one AI-generated headshot styled for professional use. No account, no card, no wait. The result comes back in seconds cropped four ways for the platforms people actually use: LinkedIn, Instagram, resume, email avatar.
That's it. Deliberately.
We had conversations about showing more two styles, multiple backgrounds, a picker. We kept pulling it back. More output meant more satisfaction. More satisfaction meant less reason to explore the paid product. Counterintuitive but true: the free tool's job was to create desire, not fulfil it.
The backgrounds were borrowed, not built
We sourced backgrounds from Freepik and blended them into the output. Not photorealistic studio quality. Polished enough to look intentional, imperfect enough that you'd never actually use it as your LinkedIn photo. That gap between "this looks great" and "but I wouldn't post this" was exactly where the paid product lived.
Resource constraints accidentally produced the right outcome. We didn't have the budget to build a perfect free tool, and it turned out a perfect free tool would have been the wrong product anyway.
The watermark was a branding decision, not a paywall
Most products watermark to protect conversion. We watermarked to distribute the brand. Every person who downloaded that free headshot and used it anywhere had InstaHeadshots in the frame. The free tool wasn't just a funnel, it was a distribution mechanism. The user did the marketing.
The Thinking Behind Four Shapes
The four crops weren't arbitrary. Each one mapped to a real surface:
Square crop: LinkedIn profile picture
Circle crop: Google profile, Slack, most email clients
Portrait crop: resume and CV header
Small square: Twitter/X avatar, app icons
One headshot, four contexts. The point wasn't variety for its own sake. It was showing the user everywhere they could show up differently: and everywhere they currently weren't showing up well.
What Validated It
No A/B test. No conversion dashboard. What we have instead is quieter but just as telling.
The tool was never iterated on after launch. In a bootstrapped team that moves fast, that means one thing: it worked. You don't leave something alone unless it's pulling its weight.
The main product it fed grew to 300,000+ professionals and 30M+ headshots generated. The free tool's job was to start that chain. The chain ran.
And then the competitors noticed.
HeadshotPro launched their own free generator — 9 headshots, one free HD download, rest locked behind payment. Their positioning copy now reads: "Other free AI tools cap you at one watermarked photo and stop there." A direct reference to the exact model we built.
Aragon followed with their own free tier — but asked for 6 photos upfront and required account creation. More friction, not less. Both products validated the category. Neither matched the simplicity.
When the two largest players in your space build versions of something you shipped first, the decision was right. That's the closest thing to a benchmark you get without internal data.
What I'd Do Differently
One thing. The free tool was designed, launched, and left. There was never a structured way to understand how many free users became paid users, or which part of the output tipped them over.
That data would have been gold — not to change the tool, but to understand it. The instinct was right. The evidence was circumstantial. Next time, I'd instrument it from day one, even just a UTM tag from the free tool to the paid signup flow.
Knowing you were right and being able to prove it are two very different things in design.