How much funding has Bespoke Labs raised?

Bespoke Labs has raised $47.25M in total funding. Their latest funding round was Series A. Key investors include Wing VC, 8VC, Mayfield, The House Fund, Jeff Dean, Tristan Handy (dbt Labs CEO).

Where is Bespoke Labs headquartered?

Bespoke Labs is headquartered in Mountain View, California, United States. The company has a team of 11-50 people.

When was Bespoke Labs founded?

Bespoke Labs was founded in 2024. It is based in Mountain View, California, United States.

How much funding has Bespoke Labs raised?

Bespoke Labs has raised $47.25M in total funding. Key investors include Wing VC, 8VC, Mayfield, The House Fund, Jeff Dean, Tristan Handy (dbt Labs CEO).

Who is the CEO of Bespoke Labs?

Maheswaran (Mahesh) Sathiamoorthy is the CEO of Bespoke Labs. The company was founded in 2024.

Where is Bespoke Labs located?

Bespoke Labs is headquartered in Mountain View, California, United States.

What funding stage is Bespoke Labs at?

Bespoke Labs is currently at the Growth Stage stage.

How many employees does Bespoke Labs have?

Bespoke Labs has approximately 11-50 employees.

Bespoke Labs

Category: AI Infrastructure

Bespoke Labs is an applied AI research lab building company-scale RL environments and infrastructure for training, evaluating, and optimizing reliable AI agents for production. Bespoke Labs was founded in 2024. The company is led by Maheswaran (Mahesh) Sathiamoorthy. Based in Mountain View, California, United States. Team size: 11-50. Total funding raised: $47.25M. Latest round: Series A. Key investors include Wing VC, 8VC, Mayfield, The House Fund, Jeff Dean, Tristan Handy (dbt Labs CEO).

Founded: 2024
Headquarters: Mountain View, California, United States
Team size: 11-50
Total funding: $47.25M

Value proposition

Bespoke Labs builds realistic, company-scale RL environments and data curation infrastructure that enable frontier labs and enterprises to train, evaluate, and optimize AI agents for long-horizon, production-grade tasks — solving the reliability bottleneck that prevents agents from moving from demo to deployment.

Products and solutions

RL environments & infrastructure (company-scale simulations with real codebases and microservices), GEPA (Genetic-Pareto Agent Optimizer for automated prompt/policy search), Curator (synthetic data curation library), OpenThoughts (open reasoning dataset), Terminal-Bench (environment-based benchmark for agentic systems), Evalchemy (evaluation/benchmark tooling), Bespoke-Stratos/OpenThinker reasoning models, Bespoke-MiniCheck (factuality model)

Unique value

Research-first approach with open-source contributions (OpenThoughts, Terminal-Bench, GEPA cited by Anthropic, OpenAI, Google DeepMind); builds environments using research techniques (not contractors); co-founded by ex-Google DeepMind researcher and UC Berkeley professor; environments mirror real companies with large codebases, microservices, realistic logs, tickets, email, Slack.

Target customer

Frontier AI labs (Anthropic, OpenAI, Google DeepMind); enterprises deploying AI agents in production; AI-native companies needing reliable long-horizon agent capabilities

Industries served

AI/ML, Enterprise Software, Technology

How they differentiate

Unlike competitors who build simple app-level environments using contractors, Bespoke Labs takes a research-first approach — its team consists of research scientists and engineers who develop novel curation pipelines, synthetic environment generation methods, and real-world infrastructure snapshotting techniques. The company leads with open research (ICLR 2026 papers on OpenThoughts, Terminal-Bench, GEPA) and is cited in frontier lab release notes (Claude 4.5).

Main competitors

Surge AI, Snorkel AI, DeepTune, Mechanize

Key partnerships

DataComp community (OpenThoughts collaboration), kluster.ai (batch processing partnership for Curator), cited by Anthropic (Claude 4.5 release notes), OpenAI, and Google DeepMind (Terminal-Bench)

Notable customers

Fortune 500 enterprises (unnamed), frontier AI labs (unnamed), Thinking Machines Lab, Meta, Amazon (users of OpenThoughts dataset)

Major milestones

Founded 2024, Launched OpenThoughts (open reasoning dataset, 500K+ downloads), Terminal-Bench accepted at ICLR 2026 and cited by Anthropic, OpenAI, Google DeepMind, GEPA algorithm accepted at ICLR 2026, Raised $7.25M seed (2024), Raised $40M seed + Series A (July 2026), 200+ teams using GEPA in production, 10K+ monthly downloads of OpenThoughts on HuggingFace

Growth metrics

200+ teams using GEPA in production; 10K+ monthly downloads of OpenThoughts on HuggingFace; 500K+ total OpenThoughts downloads; 190+ public models trained on OpenThoughts; ~40-48 employees (2026)

Market positioning

Positioned as a specialist RL-environment and data-curation vendor in the "picks-and-shovels" layer of the agent AI boom — competing against both incumbents (Surge AI, Snorkel AI) and emerging vendors (DeepTune, Mechanize, Habitat, Fleet) in the rapidly filling RL-environment vendor category.

Geographic focus

United States (Mountain View, CA) with global reach through open-source releases and frontier lab partnerships

Patents and IP

Open-source releases under Apache-2.0 (Curator, SkyRL) and MIT (Verifiers); OpenThoughts/OpenThinker artifacts under Apache-2.0

About Maheswaran (Mahesh) Sathiamoorthy

Ex-Staff Research Engineer at Google DeepMind; B.Tech from IIT Kharagpur; MS/PhD from USC; expert in large language models and recommender systems

Official website: https://bespokelabs.ai