What Is an AI Managed Portfolio?

An AI managed portfolio is an account where an algorithm, rather than you, decides and executes the trades. The word the whole thing turns on is discretion. Discretion means authority to act without asking first. When you open a managed account you grant it, deliberately, usually in the paperwork you sign at onboarding.

That is why a portfolio being visible to an AI does not make it managed. An assistant can read every position you hold, tell you that one stock is 40% of your equity, and still have no way to do anything about it. Reading and acting are different permissions, and only the second one makes something managed.

The regulator's language is useful here because it predates the marketing. The SEC's investor education material defines a robo-adviser as an automated digital investment advisory program that collects your goals, horizon, income, assets and risk tolerance through an online questionnaire, and then creates and manages a portfolio for you. That is the structure. “AI managed portfolio” is a newer name for it.

This is the part most explanations skip, and it is worth being direct about. The core of a mainstream robo-advisor is not a learning model choosing securities. It is rules-based allocation and rebalancing derived from portfolio theory that long predates machine learning: a questionnaire assigns you a risk tier, the tier maps to a mix of low-cost index funds, and software keeps you at that mix as prices drift.

That is a genuinely good product and a genuinely automated one. It is just not what most people picture when they read “AI”. Where machine learning does show up in these services it tends to be in the unglamorous places: choosing which tax lots to sell, managing cash, personalising onboarding, flagging accounts that have drifted. Those are real and they are not stock picking.

The practical consequence is that “AI” on its own tells you almost nothing about a managed product. Two services can both use the term and differ completely in what the software actually decides. The questions that do discriminate are about authority, cost and custody, which is what the rest of this page is about.

The category has a regulatory definition that is more stable than the marketing, and two of these documents are worth reading before you hand anyone discretion.

• Investor.gov (SEC Office of Investor Education and Advocacy), Robo-Adviser. Read it. Defines a robo-adviser as an automated digital investment advisory program that collects your goals, horizon, income, assets and risk tolerance through an online questionnaire, then creates and manages a portfolio for you, often at lower cost than a traditional advisory programme.
• SEC, Staff Bulletin: Standards of Conduct for Broker-Dealers and Investment Advisers, Care Obligations. Read it. Sets out the care obligations that apply to broker-dealers under Regulation Best Interest and to investment advisers under the fiduciary standard of the Investment Advisers Act of 1940. Both are drawn from fiduciary principles that include acting in the retail investor's best interest and not placing the firm's interests ahead of the investor's.
• FINRA, with the SEC and NASAA, Artificial Intelligence (AI) and Investment Fraud. Read it. A joint investor alert, published January 2024, warning that bad actors exploit the popularity of AI to lure victims, and that claims about proprietary AI producing guaranteed or outsized returns are a recognised fraud pattern.

The useful test that comes out of these: a provider managing your money is normally a registered investment adviser and therefore owes you a duty of care, which is a checkable fact rather than a claim. And the joint regulator alert is specifically about AI-flavoured return claims, so treat any promise of guaranteed or reliably market-beating performance from proprietary AI as the warning sign they say it is.

Same four fields for each, so the differences are visible rather than buried in tone: who can trade, where the money lives, what the AI is really doing, and whether the phrase honestly applies.

Most explanations stop at “an algorithm manages it”, which is the part that tells you nothing. Here is the mechanism the mainstream services actually run, because once you can see it you can judge whether it is worth what it costs.

• 1. A questionnaire turns you into a risk number. Age, horizon, income, existing assets, and a handful of questions about how you would react to a fall. The output is a score that maps to one of a small number of model portfolios, often somewhere between five and ten of them. Two people with quite different lives can land in the same tier, which is the first place personalisation is thinner than it looks.
• 2. The tier maps to a target allocation. A specific percentage in stocks, bonds, and sometimes cash or alternatives, filled with low-cost index funds chosen by the provider. This is the portfolio. Nothing about it is chosen for you individually; you were sorted into it.
• 3. Money is invested to hit the target. Deposits are bought into whichever holdings are furthest below target, which is why regular contributions do a lot of the rebalancing work on their own and why the software needs to trade less than you might expect.
• 4. Drift is monitored against a tolerance band. Prices move, so the mix moves. Most services do not rebalance on a calendar; they rebalance when a holding drifts more than a set distance from its target, commonly a few percentage points. That threshold is the single most consequential setting in the whole system and it is rarely disclosed precisely.
• 5. Rebalancing sells the winners and buys the laggards. Mechanically, staying at target means trimming whatever has run and topping up whatever has not. This is the discipline people are paying for, and it feels wrong at exactly the moments it matters.
• 6. In taxable accounts, tax rules sit on top. Which specific lots get sold, whether a loss is harvested to offset gains, and whether a purchase would trigger a wash sale. This is genuinely sophisticated software and it is where the strongest version of the “AI” claim actually lives.

Read that list back and notice what is not in it: nothing decided which companies to own. The securities came from index funds, and the funds came from a menu the provider set. The automation is in the allocation, the discipline and the tax handling, and those are the three things it is fair to pay for.

Worth thinking about before you need to know, because it is the situation that decides whether delegation was the right call for you specifically.

A managed portfolio does not get out of the way of a drawdown. The allocation is designed to be held through one, so the software will keep you at target while the value falls, and rebalancing will move money toward whatever has fallen furthest. If you were expecting the algorithm to see it coming and go to cash, that is not what these products do, and a service that claimed otherwise would be making exactly the kind of promise the regulators warn about.

The real benefit shows up in the same moment, and it is behavioural rather than analytical. The most expensive thing most investors do is sell after a fall and buy back after a recovery. A managed account makes that harder, because acting requires you to override a system rather than simply place a trade. If you know that about yourself, that friction is worth more than the fee. If you would hold anyway, you are paying a percentage of everything you own, every year, for discipline you already have.

• It suits someone who wants the job done, has a simple enough situation to fit a model portfolio, and would honestly rather not think about rebalancing. It suits a taxable account large enough that automated tax-lot handling is worth something. And it suits anyone who knows they would panic-sell.
• It does not suit someone who wants to keep a brokerage account they already have, because managing an outside account requires a mandate most connected tools deliberately do not take. It does not suit anyone who wants to hold specific companies, since the portfolio comes from a menu. And it does not suit someone whose situation is genuinely complicated: equity compensation, a business sale, an estate with several heirs. Those want a human, not a model portfolio.

Mainstream robo-advisors commonly charge somewhere around 0.25% of assets a year, and on top of that you pay the expense ratios of the funds they hold, so the all-in number is higher than the headline. Verify current pricing with the provider, because fee schedules change and a page written last year will be out of date.

A quarter of a percent sounds like nothing, and in any single year it is. The argument against it is arithmetic over decades rather than drama: a percentage of assets is charged on the whole balance every year, including the growth, so it scales with the thing you are trying to accumulate. That is the honest case for comparing a managed service against holding the same index funds yourself, and the honest case for the managed service is that plenty of people who intend to do it themselves do not actually rebalance, and do panic-sell. Paying to be protected from that is a real service.

Watch the shape of the fee as well as the size. A flat monthly subscription is cheap on a large balance and enormous as a percentage of a small one, which is worth checking if the account you are opening is small.

• You cannot inspect a decision before it happens. This is the structural cost of discretion. You find out after, and unwinding a trade in a taxable account can create a tax bill that the original mistake did not.
• Model and allocation risk. Rules built for one set of conditions can behave poorly in another, and a widely-used allocation means a lot of people hold similar things at the same time.
• Moving money in can be taxable. If funding a managed account means selling holdings you already have, that sale is a taxable event in a taxable account. Ask whether an in-kind transfer is possible.
• Custody is a separate question from performance. Assets at a US broker-dealer are typically held in a custodial account with SIPC coverage against the failure of the firm, which protects you against the broker going under and not against investments falling. Check who custodies the assets.

These sort the whole market faster than any feature comparison, and both have factual answers.

• Who can place a trade without asking you? If the provider can, it is managed. If only you can, it is not, whatever the page says.
• Where does the money live? In an account they opened, or in the one you already had. This usually tracks the first question, but not always, so check both.

If you want the version where you keep the account, our guide to AI investing without automatic trading covers the tools that analyse without acting, and AI robo-advisor alternatives compares them against the managed option directly.

Walnut is not an AI managed portfolio and is not trying to be. It connects the brokerage account you already have, read-only by default, and answers questions about the positions actually in it. Where a broker supports trading through the connection, Walnut assembles orders for you to approve; nothing is placed until you do. Your money never moves to Walnut, and Walnut is an informational tool rather than a registered investment adviser or a fiduciary.

Which means it is the wrong answer for someone who wants the job taken off their hands. If you want a portfolio managed for you, a robo-advisor is the honest recommendation and we would rather say so than pretend otherwise.