Sequoia puts $30 million into Catalyst, a startup building AI trading agents for everyday investors

On October 8, 2026, Catalyst, a startup that builds AI trading agents for retail investors, announced a $30 million seed round led by Sequoia Capital. We read Fortune's exclusive, Sequoia's own announcement post and the one-page Wilson Sonsini deal note, which names the investors and the law firm. We did not use the product, and none of these sources publishes performance data. This article covers what was disclosed and what is still open.

The round

Per Fortune and Wilson Sonsini, the investors besides Sequoia are Jump Trading, PeakXV, Lux Capital, AntiFund, Coinbase and Premji Invest. Fortune calls the $30 million a seed round and reports that the company was cofounded in 2025 by Justin Zheng, 25, and Dylan Iskandar, 21. Some secondary outlets label the same round "Series A" or "Series Seed". The primary sources say seed, so we use seed. Neither Fortune nor Wilson Sonsini gives a valuation or deal terms.

What the agents are supposed to do

Fortune describes agents that turn natural language into a trading strategy and then execute it. Importantly, "a human in the loop has to accept," so the product as described is not a fully autonomous trader. Iskandar's stated goal is that "people will simply state an intent, and we have our agents to figure out everything end to end."

Sequoia's post gives the clearest picture of the workflow. Its example starts with a thesis in plain language, such as expecting coffee prices to rise after a dry season in Brazil. Per Sequoia, Catalyst agents then gather and analyze market signals, identify a path across platforms and venues, and help the user backtest and deploy strategies with limits and safety controls. The pitch is to start from an intent rather than from the plumbing of the market.

Sequoia partner George Robson describes the aim as moving users up the sophistication curve: exploring a thesis, understanding the instruments that could express it, optimizing cost, and executing when a trigger event happens. The firm calls the long-term ambition a universal orchestration layer for turning financial ideas into trades.

The only traction number

The one usage figure comes from the company, as relayed by Fortune: a pilot that, in the company's words, "generated hundreds of millions in trading volume over a couple of weeks." The pilot focused on power traders. Fortune adds that people were being let off the waitlist on the day of the article, and that the product is expected to be widely available in time.

Treat that as a company statement. "Hundreds of millions" is a range, not a figure. It does not say how many users traded, which venues or asset classes were involved, whether the volume was real money or notional, or how any strategies performed. Trading volume measures activity, not results. A pilot with a few heavy users can post large volume with a small user count.

What it shows, and what it doesn't

The funding tells us that large investors are willing to back agent-driven retail trading. That is a signal about where venture money is going. It is not a signal about whether the agents make anyone better at trading.

Fortune itself flags the risk. Its Term Sheet editor writes that she can imagine many ways this could go awry, given prediction markets and the normalization of crypto. Zheng's reply is that the company is "not here to build a casino," and that it is building user education tools meant to explain safety and risks in a personalized way. He also said he thinks investing safely can be as fun as sports gambling. That is an intent. No source we read shows how the education tools work or how they are measured.

Questions worth asking before trusting any agent like this

These are our checklist, not claims about Catalyst's product. Each is something the sources do not answer.

• Which instruments and venues can the agent reach, and who holds the customer's assets?
• What does the human actually approve: a plain-English summary, or the exact orders?
• Are the backtests Sequoia mentions shown with costs, and is any live track record published?
• What happens when the agent misreads an intent, for example a thesis about coffee that becomes a leveraged position?
• Which regulator oversees the product, and which disclosures are shown to users?

The bigger trend

Retail brokers are moving the same way. Interactive Brokers, for example, said on its Q3 call that it invited about 3,000 clients to try autonomous AI trading, and we covered that in our IBKR article. Catalyst adds a venture-funded startup to the same trend: the interface to markets is shifting from order tickets to sentences.

For readers, the practical point is that funding and pilot volume are marketing-grade evidence. The evidence that would matter is a published, audited record of what these agents did with real accounts, including losses. Until it exists, a headline round is a reason to watch the product, not a reason to trust it. This is information, not a recommendation.