Latest / Future of Work Tech with Fexingo: Remote Tools, AI Productivity, and Workplace Software / How AI Is Automating Your Headshot Photography
Transcript
- Lucas: So you've got a new job, it's your first week, and HR sends you an email with a link. Not to a photographer's booking page, but to an app where you upload three selfies and, within minutes, you get back a dozen professional headshots — different backgrounds, different outfits, all of you looking like you just stepped out of a LinkedIn photoshoot. Luna: Yeah, I've seen those. A friend at a fintech startup showed me hers — honestly, I couldn't tell it wasn't a real photo. She said it took two minutes. Lucas: Right. And it's not just startups anymore. We're seeing adoption across the Fortune 500 — I've seen estimates that nearly 15 percent of large companies now offer ai generated headshots as an employee perk or as part of onboarding. It saves thousands of dollars per photoshoot and eliminates scheduling headaches. Luna: Fifteen percent? That's higher than I expected. Who's the main player here? Lucas: The biggest name is probably HeadshotPro — they've raised serious venture funding and claim to have served over a million users. There's also Remini, which started as a photo-enhancement app and now does headshot generation. And a bunch of smaller players like Try It On AI and Portrait Studio. Luna: And the tech underneath — it's generative adversarial networks, right? The GAN approach where one network creates the image and another critiques it? Lucas: Exactly. The generator network takes your selfie and tries to produce a realistic headshot. The discriminator network tries to spot fakes. They train against each other until the generator is good enough to fool the discriminator — and a human. The result is a high-resolution image that looks like it was taken with a DSLR and professional lighting. Luna: So the model learns what a 'good headshot' looks like from thousands of real photographer-taken portraits. But that training data has to be carefully curated, otherwise you get bias. Lucas: That's the critical issue. If the training set is mostly white men in suits, the AI will struggle with diverse skin tones, hair textures, or cultural attire. There have been well-documented cases where AI headshot tools lightened skin or failed to render afro-textured hair accurately. It's a real concern for corporate adoption. Luna: And companies are starting to audit for that. I know a few HR tech consultants who now include AI headshot bias testing in their vendor evaluations. Lucas: Good. Because the upside is real — consistent branding, faster onboarding, and honestly, a lot of people just feel more comfortable taking selfies than sitting for a formal shoot. But if the tool doesn't represent you well, it can backfire. Luna: What about the legal side? Do companies own the generated image? The employee? Lucas: It's murky. Most terms of service grant the company a license to use the generated images for internal purposes, but the model itself was trained on potentially copyrighted photos. There's an ongoing class action around training data for generative AI that could set precedent. Luna: So the employee might not actually own their own AI headshot. Lucas: Depends on the contract. HeadshotPro, for example, gives the user full commercial rights to download and use their generated images. But the model weights — the AI itself — remains proprietary. So if you leave the company, you can take the image, but you can't recreate it with their tool again without a new account. Luna: Makes sense. I want to talk about the quality leap. I remember trying a similar tool two years ago and the results were… uncanny valley. Hands looked weird, teeth were smudged. Lucas: Yeah, the progress has been dramatic. In 2023, GANs were good but often produced artifacts — strange skin textures, mismatched eye directions. Now with diffusion models and better training pipelines, the output is often indistinguishable from a real photo, even under scrutiny. Luna: So much so that some companies are using these for official ID badges. That's a big trust signal. Lucas: I've seen that too. And it raises an interesting question: If your official corporate photo is ai generated, what does that do to personal branding? Your LinkedIn profile becomes a synthetic representation. Is that dishonest, or just efficient? Luna: I think it's efficient, as long as it looks like you. If it's a realistic likeness, it's no different than using a filtered photo. The problem is when the AI 'enhances' you — makes you look thinner, younger, more symmetrical — that's where it becomes deceptive. Lucas: Exactly. And some tools offer retouching sliders. You can adjust jawline, eye size, even add a smile. That's essentially a digital plastic surgery. HR departments are starting to set guidelines: no retouching beyond what a professional photographer would do. Luna: So the tool itself isn't the problem, it's how you use it. But the tool can nudge users toward certain looks. Lucas: Right. The default outputs tend to favor conventional attractiveness — which is a whole other bias conversation. But the genie is out of the bottle. AI headshots are becoming standard. I checked a few weeks ago and on Upwork, 'AI headshot generation' is one of the fastest-growing freelance skills. Luna: Wait, freelancers are offering this as a service? So you don't even need to use the tool yourself — you send your selfies to someone who runs them through a model and delivers the finals. Lucas: Exactly. It's a micro-service. Charge twenty bucks, take ten minutes, deliver twenty headshots. The freelancer handles the lighting and pose selection. It's a whole new category of virtual assistant work. Luna: That's fascinating. It also means the barrier to a professional online presence is basically zero now. Anyone with a smartphone can have a studio-quality portrait. Lucas: Which is great for job seekers, but it also means the headshot arms race escalates. If everyone has a perfect photo, the photo itself becomes less differentiating. You have to rely on actual substance. Luna: Maybe that's the hidden upside — we stop judging by appearance and focus on skills. But I'm not that optimistic. Lucas: Probably wise. But let's stay concrete: What's the practical takeaway for a listener whose company is considering an AI headshot tool? Luna: Test it on a diverse group first. Check for skin-tone accuracy across the Fitzpatrick scale. And make sure the terms of service let you keep your image. Lucas: And if you're an employee offered one, take the selfie in good lighting, plain background, neutral expression. The better the input, the better the output. Garbage in, garbage out still applies. Luna: Speaking of practical — if today's tech conversation gave you something usable, it's worth mentioning that these episodes are ad-free and listener-supported. A couple of bucks a month genuinely keeps the lights on. If you've gotten value, buy me a coffee dot com slash fexingo. Small amounts, real difference. Lucas: Yeah, it's how we stay independent and avoid the sponsored-segment pressure. So if you can chip in, thanks. And we'll keep drilling into the tech that's actually reshaping work. Luna: Alright, back to headshots. One thing we haven't touched on: What about video? If AI can do a headshot from a selfie, can it generate a whole video avatar for virtual meetings? Lucas: It can, and it's already happening. Companies like Synthesia and HeyGen let you create a talking-head video from a single photo. But that's a whole other episode. For now, the headshot wave is here, and it's probably on your company's radar within the next year. Luna: Something to smile about — even if the smile is ai generated. Lucas: Exactly. Keep the input real, and the output will serve you well.