Gaia is so back: 5,000 users, 600K+ renders, what's next

Over a year ago I paused Gaia. I went back to school — maybe because I had more questions than answers, and I couldn't build the right tool without first understanding the problem more deeply. We really wanted to help the industry design better buildings, faster. We built the first version at Cornell together — incredible journey, real users, real lessons. But the gap between 'AI tool for architects' and 'tool that actually understands architectural thinking' kept nagging at me.

So I'm now a PhD researcher at the Technion, doing Computer Science + Human-Computer Interaction research at the Architecture faculty — studying how AI can actually serve the way architects design, not just produce sexy renders. The doctorate gave me permission to close that gap properly — not on a fundraising timeline, but on a research one.

5,000 users. 600,000+ images.

The numbers surprised me when I looked at them. Half a million images generated by practicing architects, students, and designers using Gaia as a real part of their workflow. That's not a toy. That's infrastructure. The best times are genuinely ahead.

What V2 ships

• Generate: prompt to photorealistic render with full camera, light, and material control
• Edit: brush-led masking — change specific regions without touching the rest
• Research: site analysis, zoning, and precedents in conversation with Gaia
• Interior AI: virtual staging for real estate in 20 seconds
• Knowledge graph: your entire design history, visualized

The long game is the intelligence layer — the twin that learns your vocabulary, your jurisdictions, your way of seeing a building. V2 gets us closer to earning that trust. The intelligence isn't 'AI-powered' as a marketing claim — it's grounded in what computer science and architecture research are actually producing right now, fed back into how the tool thinks.

Feedback? Reply to any email, or find me at daan@gaia.computer. I read everything.