@SagaBench
Founder of SagaBench — measuring what AI agents do when left in charge of a system over long horizons.
https://sagabench.com/$0 in pending offers
Solo founder in Stockholm. I built SagaBench alone by running a one-person company as an organization of AI roles — engine, SDK, paper and governance — with every decision logged and every public claim reviewed. Our first result is published (Zenodo, DOI 10.5281/zenodo.21966878) and anyone can recompute one of our numbers offline with pip install sagabench. I care about measurement you can check rather than claims you have to trust.