Robinhood has built AI agents that trade for customers inside its own app, and Vlad Tenev, its co-founder and CEO, told Alex Heath on the Sources Podcast that more than 100,000 customers have already set one up.
Most of the debate about AI and money is about whether the machines can be trusted. Vlad spent much of the hour on a different question: who gets to own the upside.
"We want to put all of the power of a sophisticated hedge fund or algorithmic trading firm in your pocket." — Vlad Tenev
Vlad grew up in Bulgaria during its early-1990s hyperinflation, co-founded Robinhood, and also co-founded Harmonic, an AI mathematics company. He is now pushing Robinhood's customers into agentic trading, private AI companies and tokenized shares at once.
The full interview is covered here so you can skip it. 60 minutes of audio, 8 minutes of reading.
Here are the 7 takeaways that matter.
Sources Podcast · Oct 13, 2026 · 1h · 52m saved
With Vlad Tenev, Co-Founder and CEO of Robinhood
Key Takeaways
Robinhood keeps agent trading in a separate account that customers fund on purpose, and every trade needs approval by default
Vlad's long-term goal is for retail customers to beat hedge funds, eventually with access to low-latency infrastructure at the exchanges
Edge comes from different data and new markets, not from the model, because an AI everyone has gives no one an advantage
Vlad blames much of the AI backlash on who owns AI, not on the technology
Robinhood Ventures is his answer: listed funds that hold private AI companies before they go public
Robinhood is shipping tokenized stocks abroad first to prove the product before asking US regulators
Vlad argues against AI labs keeping their strongest models for internal research and wants outside scientists using them
1. Agents Built Into the App
Robinhood Agents grew out of an earlier release, Robinhood for agents, an MCP connection that let customers plug tools such as Claude Code or Codex into a trading account and an agentic credit card. Vlad said that version had two problems. Most customers do not want to set up developer tools, and general-purpose models are often reluctant to trade because their training data holds little trading behavior.
Robinhood moved agent setup into its own app and tuned the models to use its tools
Setup now takes about 60 seconds: name the agent, pick a model, and decide whether trades need approval. The product launched with cash-settled stocks only and has since added options, crypto and limited margin. Vlad said more than 100,000 customers have onboarded, and retention improves each time a capability such as margin is unlocked.
Robinhood is building the category and the product at the same time
"It's sort of like we're wandering through the fog and iterating on the infrastructure and the product at the same time." — Vlad Tenev
Robinhood has added capabilities in stages, from stocks to options, crypto and margin, while it works out the infrastructure underneath.
