Retail

Podium Markets AI built Ivy to read a retail investor's entire portfolio across every brokerage they hold an account at, then recommend trades, but deliberately stopped short of letting it place them, on the explicit bet that investors want the analysis without giving up the decision

CEO Dirk Mueller-Ingrand frames it as "the AI informs, but the human decides," a narrower and more cautious design than the fully autonomous execution agents Robinhood, Public, Coinbase and eToro have all shipped this year, betting that trust in cross-account AI recommendations has to be earned before it is automated away.

· Source: CNBC / Podium Markets AI


Every other brokerage AI story this year has been a race toward the same finish line: agents that execute trades autonomously, around the clock, with the human increasingly relegated to setting goals rather than approving actions. Podium Markets is running in the opposite direction on purpose. Ivy aggregates a user's holdings across every brokerage account they connect, which is itself the harder engineering problem, most retail AI tools only see one account, and generates goal-aligned recommendations. Then it stops. No execution, no standing authorization, every trade requires the human to actually click approve.

The cross-brokerage aggregation is the part worth taking seriously as a product decision, not the AI layer on top of it. Most retail investors do not hold their entire portfolio at one broker, they have a 401k somewhere, a taxable account somewhere else, maybe a crypto position on a third platform, and no single institution has ever had a complete picture of the whole thing to advise against. An AI assistant that can see across all of it and reason about the portfolio as a single object, rather than optimizing one account in isolation, is solving a real gap that predates AI entirely. The novelty is that GenAI made the natural-language interface to that aggregated view cheap enough to build.

The withheld-execution design is a bet on where retail trust currently sits, and it is a reasonable one. Robinhood and Public letting third-party or proprietary agents trade directly is a bigger claim on user confidence than most retail investors have been asked to extend to a piece of software before, and the industry does not yet have a multi-year track record of autonomous retail agents performing well through a real drawdown. Podium's founder framing, that the AI informs and the human decides, reads less like a permanent philosophical stance and more like a calibrated first step: earn trust on the recommendation quality first, and the execution question can be revisited once there is a track record to point to.

The risk in the cautious design is competitive, not technical. If Robinhood, Public, and the brokerages racing toward full execution build a large enough usage and volume advantage while users get comfortable with autonomous agents elsewhere, the human-in-the-loop model could end up looking conservative rather than trustworthy by comparison, a nice-to-have research layer next to platforms that actually do the work. Podium's advantage has to come from being genuinely better at the cross-account analysis than what the execution-focused platforms bolt on as an afterthought, because analysis quality is the only lever left once the market normalizes around agents that can also pull the trigger.


Read the original: CNBC / Podium Markets AI - Podium Markets AI built Ivy to read a retail investor's entire portfolio across every brokerage they hold an account at, then recommend trades, but deliberately stopped short of letting it place them, on the explicit bet that investors want the analysis without giving up the decision. Commentary is the independent editorial view of Share Trading; the original article is credited to its publisher.