Photo AI is a consumer AI photography platform built by Dutch indie founder Pieter Levels (@levelsio). You upload 5–20 photos of yourself, it trains a private DreamBooth-style model of your face in about a minute, and then you generate photorealistic photos of yourself in any setting, outfit, or pose by writing a prompt or running a pre-made \"photo pack\" (Tinder, LinkedIn headshots, luxury lifestyle, film noir, etc.). Beyond avatars, it has expanded into video (image-to-video, talking videos, motion capture, a browser-based multi-track video editor), clothing try-on (including batch try-on for Shopify sellers), AI influencer creation (design a synthetic persona from scratch), Magic Edit (text-based photo editing), Zoom Out (uncropping), upscaling, photo restoration, and even 3D model export (.GLB). It also offers an MCP connector so you can ask Claude or ChatGPT to \"take 4 photos of me in Paris\" and have them appear in the chat. The product is 100% independently owned, not VC-funded, and run by Pieter with no employees. Pricing is credit-based per subscription tier (Starter, Pro, Premium, Ultra), with yearly plans over 50% cheaper than monthly. Critical caveats: the refund policy is strict—you can self-refund only if you have not yet created a model and have generated fewer than 20 photos, and referral signups are not eligible; the 3D model feature is explicitly noted by the founder as not great yet, especially for people; and the AI influencer monetization examples on the page (Lil Miquela, Noonoouri, etc.) are aspirational, not typical outcomes.

Pros & Cons:
Pros:
The training pipeline is genuinely fast—about a minute for a personal model, versus 25–50 hours on a local laptop, which removes the biggest technical barrier to personalized AI photos; the breadth is unusual for an indie product: avatars, try-on, video, talking video, mocap, 3D export, Magic Edit, and Zoom Out all in one subscription, so you don\'t need separate tools for each; the MCP connector lets you drive the tool from Claude or ChatGPT in natural language, which is a forward-looking workflow for users already living in AI chat; photo packs are concrete and organized by use case (dating, professional, creative), which is more actionable than a blank prompt box; the founder is transparent about limits (3D quality, refund policy, pricing structure) rather than overselling; and the affiliate program gives 1,000 credits to both parties with a 30-day cookie, which is more generous than most referral programs.
Cons: The refund policy is restrictive—no refund once you\'ve created a model or generated 20+ photos, and referral signups are excluded, so evaluation carries real financial risk if you train a model and then find the output unsatisfactory; the credit system is granular and easy to underestimate—a video costs 300 credits, a mocap video 50, hyper-realism photos 5 each—so cost can add up faster than the subscription price suggests; the 3D model output is explicitly low quality for people per the founder\'s own note, so it should not be treated as a production 3D tool; the AI influencer revenue examples are marketing framing, not realistic expectations for most users; and the product is a solo-founder operation with no employees, so support depth and continuity depend entirely on one person.

Who is it for?:Individuals who want high-quality, photorealistic AI photos of themselves for dating profiles, social media, or professional headshots without hiring a photographer; creators and e-commerce sellers who need model try-on photos or batch content for Shopify-style stores; users curious about AI influencers or video generation who want one subscription covering photos and video; developers or AI-native users who want to drive photo generation from Claude/ChatGPT via MCP. Not suited for anyone who needs a refund guarantee after training a model (the policy is strict), teams needing enterprise support or SLAs (solo founder, no employees), users expecting production-grade 3D models of people, or buyers who require independent third-party review volume before committing.