Public Launches AI Agents That Trade on Prediction-Market Signals

A shift in the odds of a rate cut can now prompt more than a notification inside Public. The investing app has launched AI Agents for Prediction Markets, letting members trade event contracts or use those markets as signals for trades elsewhere in a portfolio. Investors approve the rules first; after that, an agent can act when the conditions are met.

An event market becomes a portfolio signal

Public partnered with Kalshi to add prediction markets to its platform. Kalshi supplies the event contracts and handles trades in those markets, while Public gives members a place to access them and set up agents. Public says Prediction Markets are available to all its members.

The markets cover subjects including economics, corporate events, commodities, climate and politics. Members can take a position on an event itself. The more distinctive use is to watch a market’s changing probability and make it a trigger for a stock or bond trade. In that case, the investor is not necessarily trading the event contract; its price is informing a separate portfolio decision.

Approval happens before the signal arrives

A member describes a strategy to Public’s AI. The chatbot turns that request into a visual plan for the member to review, then into fixed rules that require approval before they run. Those rules might call for an alert when odds move or for an automatic purchase or sale. Once activated, Public says, an agent cannot trade outside the conditions its customer approved.

That sequence places the human decision at setup, not necessarily at the moment an order is placed. Public cofounder and co-CEO Leif Abraham told Fortune the agents are meant to monitor markets and execute strategies when investors are not watching a screen. The tradeoff is direct: approval can constrain what an agent does, but a member choosing automatic execution is still authorizing a future trade before seeing the market move that triggers it.

The choices behind a probability threshold

Public’s examples show how much discretion remains with the investor. One proposed rule buys shares if a market-implied probability of regulatory approval crosses a chosen threshold. Another only sends an alert when expectations for rate cuts jump, along with a summary of the member’s bank-stock exposure. A third would buy put options, which can gain value when a stock falls, if the market assigns a sufficiently high probability to an earnings miss.

These are examples of instructions members could give, not evidence that the strategies make good trades. They also illustrate decisions the agent does not settle for its user: which market to follow, what change should count as meaningful, and whether the response should be an alert or an order. Public’s approval step governs the chosen rule. It does not make the event market’s probability a certain forecast.