Jigsaw
Applied research lab scaling RL environments to accelerate superintelligence.
About
Jigsaw is an applied research lab automating the creation of RL environments. We are training foundation models and building infrastructure for world generation, expert authoring, and evaluation, enabling high-fidelity environments and the transfer of domain expertise into model capability at scale. Jigsaw provides the infrastructure to support the shift toward sovereign-owned intelligence.
Founders · 2
Prior to Jigsaw, Miguel was a researcher at Stanford AI Lab working on agents and evals. In 2024, he dropped out of university and moved to Tokyo to build simulated worlds for agents ($200k+ MRR). Before that, Miguel worked as an AI engineer in big tech and top research labs.
Prior to Jigsaw, I cofounded a startup which focused on building agentic environments situated around games. Scaled that to 6 figs monthly revenue. Before that I scaled enterprise partnerships at Polygon Labs. Graduated early from USC in 2022.
Related startups
Data-centric infrastructure to accelerate the development of AI
Agent simulation and RL for researchers
