This startup builds 'flight simulators' for AI agents — just raised $43M
Deeptune raised $43M from a16z to build virtual training environments where AI agents practice real workplace tasks — like flight simulators, but for office work.
Deeptune just raised $43 million in Series A funding led by Andreessen Horowitz to build something AI desperately needs: practice environments. Think of it as flight simulators — but instead of training pilots, they're training AI agents to do office work.
Why AI agents need a gym before they get a job
Here's the problem: most AI today learned by reading the internet. That's great for answering questions, but terrible for doing things. If you want an AI agent to file an expense report in Salesforce, respond to a customer ticket in Zendesk, or build a financial model in Excel — it needs to practice those specific tasks.
That's exactly what Deeptune builds. The company creates pixel-perfect replicas of real workplace software — Slack, Salesforce, ticketing systems, finance tools — where AI agents can try, fail, and learn without breaking anything real.
"You wouldn't have a pilot who has only ever read books or watched tutorials fly a plane. You would put them in a flight simulator." — Tim Lupo, Deeptune CEO
Who's backing the bet — and why $43M
The round was led by Andreessen Horowitz (a16z), one of Silicon Valley's most influential venture firms. Other investors include 776 (Alexis Ohanian's fund), Abstract Ventures, and Inspired Capital.
The angel investor list reads like an AI all-star roster:
- Noam Brown — OpenAI researcher behind some of the company's most advanced reasoning systems
- Brendan Foody — CEO of Mercor, a fast-growing AI hiring platform
- Yash Patil — CEO of Applied Compute
The team of roughly 20 people, based in New York, includes former employees from Anthropic, Scale AI, Palantir, Hebbia, Glean, and Retool — a who's who of AI infrastructure companies.
What AI agents actually practice
Deeptune's training environments simulate real job tasks across multiple industries:
Customer support — handling tickets across multiple tools
DevOps — monitoring systems, responding to alerts
Video editing — navigating editing software workflows
Finance — building leveraged buyout (LBO) models in Excel
The company says it has already built hundreds of training environments for leading AI labs, and its work has "contributed to recent advances" in AI agents' ability to use computers — a capability that companies like Anthropic and OpenAI have been racing to improve.
A $90 billion market is forming
The global reinforcement learning market (the AI training technique Deeptune uses) was worth $11.6 billion in 2025 and is projected to reach $90+ billion by 2034. Major AI labs are reportedly considering spending more than $1 billion on training environments alone.
As a16z partner Marco Mascorro put it: "Deeptune has built a platform enabling this shift, allowing top labs to train and evaluate behaviors in reliable, scalable ways."
Why this matters for everyone — not just developers
If you've ever been frustrated by an AI assistant that can answer trivia but can't actually do anything useful in your work tools, Deeptune is working on the fix. The better AI agents get at practicing in simulated environments, the sooner they'll be able to reliably handle real tasks — filing reports, scheduling meetings, processing data — without making costly mistakes.
The era of AI that just talks is ending. The era of AI that actually works is being trained right now — in virtual gyms built by companies like Deeptune.
Related Content — Get Started with Easy Claude Code | Free Learning Guides | More AI News
Stay updated on AI news
Simple explanations of the latest AI developments