en.Wedoany.com Reported - Mountain View-based startup Bespoke Labs has announced the completion of a $40 million funding round to build training and testing environments for AI agents. The funding includes a Series A round led by Wing VC and an early-stage seed round led by 8VC. Investors include angel investors working at Anthropic, OpenAI, and Meta, as well as Google DeepMind's Jeff Dean and dbt Labs' head Tristan Handy.

Bespoke Labs builds simulated versions of real enterprises, including large codebases, microservices, logs, support tickets, emails, and Slack threads. Agents train in these virtual worlds, learning long, multi-step workflows. The company also helps customers measure and tune agents, using its internal optimizer GEPA to find better prompts and strategies more quickly, rather than through manual tuning.

Founded in 2024 by CEO Mahesh Sathiamoorthy and Chief Scientist Alex Dimakis, the company has a team of about 40 people with an academic-oriented culture. Bespoke Labs is a core contributor to Terminal-Bench, a widely cited benchmark for agent skills. The company also built OpenThoughts, an open reasoning dataset that has been downloaded over 500,000 times by labs including Meta and Amazon. Bespoke treats environment building as research, rather than outsourcing the work to contractors, and sells the resulting infrastructure.

Independent tests show that the length of tasks agents can reliably complete roughly doubles every seven months, with some analyses suggesting this cycle is closer to every four months. Bespoke Labs' competitors include companies focused on self-learning agents, those conducting stress tests and benchmarking before release, and those chasing the economics of running agents at scale. Bespoke is betting on the training ground, not the model.