Fundraise

Thrixel raises $3M to build 3D assets AI agents can actually use

What's the deal? ThrixelDealroom has a profile for this one. Try Dealroom →, a Chicago-based developer of a 3D generation platform built for AI agents, has raised $3 million in seed funding. Sierra Ventures, MFV PartnersDealroom has a profile for this one. Try Dealroom →, Harper Court VenturesDealroom has a profile for this one. Try Dealroom →, and Z5 CapitalDealroom has a profile for this one. Try Dealroom → backed the round, alongside AI and computer graphics researcher Jon BarronDealroom has a profile for this one. Try Dealroom → and former EA executive Matt BilbeyDealroom has a profile for this one. Try Dealroom →.

What's the endgame? Founded by University of Chicago computer scientist Rana Hanocka, Thrixel generates 3D assets built from individually named parts with logical hierarchy. That structure lets AI agents identify, assemble, and animate components directly inside industry-standard game engines.

How does it work? The platform combines vision language models, agentic pipelines, generative diffusion, and optimisation techniques, each mapped to a step in the 3D modelling pipeline. Creators can run the full workflow or apply specific steps, and manipulate individual parts directly with code.

What's new? Alongside the platform, Thrixel is launching Build World, an agentic workflow that turns ideas into interactive 3D experiences. Creators work with an AI coding agent that generates 3D content while writing the underlying logic, with support for Roblox, Three.jsDealroom has a profile for this one. Try Dealroom →, UnityDealroom has a profile for this one. Try Dealroom →, and Unreal EngineDealroom has a profile for this one. Try Dealroom →.

Why now? Thrixel is courting partners across game development, interactive media, robotics, and simulation — sectors racing to feed AI pipelines with usable, structured 3D content.

The signal: At $3 million, the round sits in the 34th percentile for deal size, a modest cheque for an early-stage bet. But the part-aware approach targets a real gap: most generative 3D output is a single mesh AI agents cannot easily edit or reason about.

Read more: finsmes.com

Image credit: Thrixel

More top stories