Your Next ETF May Be Picked Entirely By An AI

What The Filing Actually Says

BAILA does two jobs, according to the prospectus. It reads the macro weather, sorting the market into “risk-on” or “risk-off” by measuring the present against more than three decades of history, and it builds the portfolio, sizing each holding by what it expects to earn against the risk it carries. The flagship fund aims to beat the S&P 500 over a full market cycle with a tight book of 20 to 40 US stocks, drawn from the 1,000 most heavily traded names on the market. In a calm market it leans into momentum and runs its stock exposure toward full. When the model smells trouble, it can retreat into defensive shares, dial down the portfolio’s sensitivity to market swings, or sit entirely in cash.

Quantitative funds have read patterns and shifted exposure for decades, so the method is not the novelty. The novelty is that the marketing pitch is the machine. A passage headed “the Utilization of AI in the Fund’s Investment Process” promises that AI lets the adviser “analyze vast amounts of financial data, market trends, and economic indicators in real-time,” spot opportunities and risks “with greater accuracy and speed,” and keep “a well-balanced and high-performing portfolio.” That is the cadence of a brochure, and it is worth marking where in the document it appears.

The Sales Page And The Risk Page

A few pages on, the confident voice deserts the document. The risk section concedes that the model’s success “depends on the quality, reliability, and timeliness of the data input,” and that “if data is stale, missing, or otherwise inaccurate, the Fund’s investment decisions may be adversely affected.” It grants that the adviser leans on third-party data, and that if a supplier cuts off access or feeds it bad numbers, the strategy “may be impaired.” Then the clincher, set down a few pages after all that talk of accuracy and high performance: “There can be no assurance that the use of AI will result in improved performance or reduced risk.”

Prospectuses are built to sell on one page and protect on the next, so a gap between the two registers is nothing new. What makes this gap worth a reader’s attention is that the thing being sold is the intelligence itself. When the pitch and the product are the same word, the space between the brochure and the boilerplate is precisely where a buyer can be led astray, and it happens to be the space a securities regulator has promised to watch.

Why The Regulator Is Watching This Word

“AI washing” is the term the SEC reached for in March 2024, when it settled charges against two investment advisers, Delphia and Global Predictions, for overstating how much artificial intelligence actually drove their work. The two firms paid civil penalties of $225,000 and $175,000. The message was blunt: claim AI, and the claim had better match the reality, or it is just an old-fashioned misrepresentation in a new coat. That enforcement record is the lens through which these three funds will be read.

To its credit, the Ai Funds prospectus includes the sentence a regulator would want to see. Despite the heavy reliance on the model, it says, “the ultimate investment decisions remain subject to oversight and approval by the Adviser’s portfolio managers.” A human is meant to sign off. That clause carries more weight than its length suggests; it marks the line between a fund run by a machine and a fund advised by one, which is also the line between an AI claim that is accurate and one that is actionable.

Who Is Actually Behind The Machine

Names matter in a prospectus, and the names here repay a careful look. The adviser across all three funds is Ai Funds, Inc. For the flagship equity fund, the sub-adviser doing the risk-management work is Milliman Financial Risk Management, and the portfolio manager is Dr. Tal Schwartz, listed as the adviser’s chief executive, chief investment officer and chief compliance officer at once. The crypto fund in the same registration runs differently, with Tuttle Capital Management as sub-adviser and a different manager, Gavin Beauregard, at the helm.

Small fund shops often share roles among a few people, and none of this is improper. It is, though, the sort of detail a careful buyer weighs. On the flagship fund, the person designing the strategy, the person investing the money and the person policing the firm’s own compliance are one and the same. For a product whose entire selling point is disciplined, rules-based decision-making, the concentration of authority at the top is worth holding in mind. Nothing here is amiss. But a spread of independent judgment is part of what protects an investor, and a prospectus is where you learn how much of it there is.

The Crypto Cousin With A Cayman Detour

The second fund in the filing, the Multi Crypto Coin AI-Managed ETF, ticker CCAI, deserves a paragraph of its own, because it carries a wrinkle the equity fund does not. It can put up to 60% of its money into cryptocurrencies, and it routes that exposure through a wholly owned subsidiary in the Cayman Islands. The prospectus states the catch in flat terms: that subsidiary “is not registered under the 1940 Act and is not subject to all the investor protections of the 1940 Act.” The same fund warns that its tax status as a regulated investment company is uncertain, and that losing it “could have adverse consequences for the Fund and its shareholders.” A buyer drawn in by the AI branding could easily miss that they are also buying an offshore crypto vehicle with thinner legal protection than a normal US fund.

What You Would Be Buying

The appeal is easy to grasp. The promise is a fund that rises with the market in good weather and steps aside before the storm, all governed by a model that has studied 30 years of history and never panics. The reality, by the document’s own account, is more hedged. The fund is brand new, with “no operating history” and no record to judge. It is non-diversified, free to pile more money into fewer names, which sharpens both the wins and the losses. Its stock exposure can sit at nought or at full, so its behaviour in a sharp fall hangs entirely on whether BAILA reads the danger in time. And the central promise, that the AI lifts returns or curbs risk, sits beside the filing’s own written warning that it may do neither.

Cost and structure round out the picture. The flagship fund charges a management fee of 0.85% a year; the crypto fund, 0.95%. As exchange-traded funds, all three trade like shares, so their market price can drift from the value of what they hold, and in stressed markets a buyer “may pay significantly more or receive significantly less than the underlying value.” The document spells this out in the standard ETF risk language every such fund must carry. None of it is concealed. All of it lives in a prospectus that, like most, far more people will buy on than will ever read.

The Old Test For A New Product

Never invest in something you cannot explain to someone else, runs the old piece of City wisdom, and an AI-managed fund tests that rule harder than most. The honest answer to “how does it choose?” is, in part, that the model weighs decades of patterns in ways its own makers describe in probabilities, not certainties. That is not a mark against these funds; plenty of sound investing rests on probabilities. It is a reason to read the document rather than the badge on the door.

Fully AI-managed ETFs reaching the registration stage is a small milestone worth marking, because it shifts algorithmic stock-picking out of hedge funds and quant desks and onto the same shelf as the index trackers in an ordinary retirement account. The Ai Funds prospectus is, in fairness, more candid than its branding: the risk pages say the quiet part out loud. The job for anyone tempted by the pitch is to read past the page that sells the intelligence and onto the page that hedges it, because the second page is where the fund tells you what it knows about its own machine, which is, in its own words, that there can be no assurance it will help at all.