Where is Deeptune headquartered?

Deeptune is headquartered in New York City, New York, USA. The company has a team of 11-50 people.

What does Deeptune do?

Deeptune operates in the AI Infrastructure sector. Developer of high-fidelity reinforcement learning simulation environments ('training gyms') where AI agents practice real-world enterprise tasks before deployment. Solves AI 'data exhaustion' by transforming data collection from a labor problem into an engineering/compute problem — generating high-quality training signals through trial-and-error in realistic digital simulations rather than scraping static web data

When was Deeptune founded?

Deeptune was founded in 2022. It is based in New York City, New York, USA.

Who is the CEO of Deeptune?

Tim Lupo is the CEO of Deeptune. The company was founded in 2022.

Where is Deeptune located?

Deeptune is headquartered in New York City, New York, USA.

What funding stage is Deeptune at?

Deeptune is currently at the Acquired stage.

How many employees does Deeptune have?

Deeptune has approximately 11-50 employees.

Deeptune

Category: AI Infrastructure

Developer of high-fidelity reinforcement learning simulation environments ('training gyms') where AI agents practice real-world enterprise tasks before deployment. Deeptune was founded in 2022. The company is led by Tim Lupo. Based in New York City, New York, USA. Team size: 11-50. Total funding raised: $46.1M. Latest round: Series A. Key investors include Andreessen Horowitz, Seven Seven Six, 776, Abstract Ventures, Inspired Capital, Pentas Ventures, Alexis Ohanian, Gary Vaynerchuk.

Founded: 2022
Headquarters: New York City, New York, USA
Team size: 11-50
Total funding: $46.1M

Value proposition

Solves AI 'data exhaustion' by transforming data collection from a labor problem into an engineering/compute problem — generating high-quality training signals through trial-and-error in realistic digital simulations rather than scraping static web data

Products and solutions

Training gyms: high-fidelity RL simulation environments replicating enterprise software (Slack, Salesforce, spreadsheets, ticketing/finance/monitoring tools) for AI agents to practice multi-step professional workflows, Prebuilt environments with problems, datasets, and infrastructure, Computer-use and code RL environments contributing to benchmarks like OSWorld and Terminal-Bench

Unique value

'Flight simulators for AI agents' — the only platform building high-fidelity, full-software replicas of enterprise environments (Slack, Salesforce, Excel, etc.) where AI agents can practice, fail, and learn through reinforcement learning at scale

Target customer

Frontier AI research labs and enterprises deploying AI agents that need to perform complex, multi-step enterprise tasks across software tools

Industries served

AI research, Enterprise software automation, DevOps, Customer support, Finance/accounting workflows

Technology advantage

Hundreds of prebuilt enterprise software simulations; tight feedback loops with frontier AI research labs; platform that generates high-quality RL training signals at scale; first to prove that 'training gyms' work for computer-use agent training

How they differentiate

First-mover in building production-grade RL environments that faithfully recreate enterprise software interfaces (not just sandboxed toy tasks). Deep integration with frontier labs' actual training pipelines. Environments already contributed to SOTA computer-use benchmarks (OSWorld, Terminal-Bench). Team of operators from Anthropic, Scale AI, Palantir, Hebbia with direct lab relationships.

Main competitors

AIChamp (custom RL environments), Braintrust (AI evaluation), Maxim AI (AI evaluation), Scale AI (expanding into RL environments), In-house lab-built environments (OpenAI, Anthropic, etc.)

Key partnerships

Andreessen Horowitz (lead investor), Worked closely with leading AI labs (frontier research labs), Mercor (customer before acquirer), Integrated with enterprise software ecosystems (Slack, Salesforce)

Notable customers

Frontier AI research labs (names not publicly disclosed but described as 'leading AI labs'), Mercor (was a customer before acquisition)

Major milestones

2022: Company founded by Tim Lupo and Lukas Schmit, 2023-03: Raised $3.1M Seed led by Seven Seven Six for AI dubbing platform, 2025-2026: Pivoted to RL training gyms for AI agents, 2026-03: Raised $43M Series A led by a16z (Fortune exclusive), 2026-07: Acquired by Mercor ($10B valuation), team joined Mercor NYC office

Market positioning

Pioneer in the emerging RL environments infrastructure category — positioned at the intersection of AI training data, simulation, and agent evaluation. The global RL market is projected to grow from $11.6B (2025) to $90B+ (2034).

Geographic focus

United States (New York City-based, serving global frontier AI labs)

About Tim Lupo

Ex-Hebbia.AI Founding Engineer; University of Southern California BS in CS and Business

Latest news about Deeptune

Deeptune raised $43M Series A led by a16z to build "training gyms" — simulation environments mimicking tools like Slack and Salesforce where AI agents learn via reinforcement learning. a16z calls RL e

More AI Infrastructure companies

Official website: https://deeptune.com