Walnut vs Fidelity Go: Which Is Better in 2026?

Figures and features are point-in-time and change; treat the table as a starting map, not a live quote.

The AI that knows your portfolio. Ask anything in plain English, research any stock or fund, and get an honest second opinion, on the broker you already use. Read-only by default, you approve every trade, and it never holds your money. Walnut is not an investment adviser.

How it works: You connect the brokerage you already use through a secure, read-only link, so your login stays with your broker and Walnut sees positions rather than credentials. You then analyze and manage that portfolio by talking through Claude or ChatGPT, and you can group holdings into thematic baskets built around a thesis. When you decide to act, Walnut prepares the trades and routes them back to your own broker for you to approve.

In practice, Walnut’s AI conversational + thematic baskets + trade. It falls under chat-driven management of your own brokerage, which makes it best suited to talking to your own brokerage with ai. On connecting an account it is “Yes (your own broker)”, and on execution it is “Read + you approve”. It is priced as free tier.

One honest limitation: It sits on top of your broker rather than being a broker itself, so you connect an existing account.

Fidelity's automated service, free below a stated balance and built from zero-expense-ratio Fidelity Flex funds. Best for people starting out.

How it works: A questionnaire sets the allocation, and Fidelity invests it in its own Flex mutual funds, which carry no expense ratio because Fidelity earns the advisory fee instead. Below a stated balance there is no advisory fee at all, which makes it one of the genuinely free ways to get a managed portfolio; above it, a flat percentage applies.

In practice, Fidelity Go’s AI automates a fidelity flex fund portfolio. It falls under hands-off automated investing (robo-advisors), which makes it best suited to small balances, where it is free. On connecting an account it is “No (holds your money at Fidelity)”, and on execution it is “Automated”. It is priced as free under a stated balance, then a flat percentage (verify current).

One honest limitation: No tax-loss harvesting, which removes the strongest argument for a managed taxable account.

The honest framing: Fidelity Go is about fully automated, hands-off investing where the platform builds and manages a diversified portfolio for you, while Walnut is about keeping your own brokerage and staying in control of what you hold, with an AI you talk to. Fidelity Go holds your money and invests it for you, whereas Walnut sits on top of the broker you already use and leaves you in control.

Every tool gives something up for what it does well. Here is the honest give-and-take on each, so you can weigh the specific strengths against the limitations that come with them rather than judging on the headline category alone.

For AI investing tools, the distinction that matters most is whether the tool works from your actual, connected positions or reasons from something else: a separate account it manages for you, or the tickers and numbers you feed it. It decides how personal the answers can be, and where your money physically lives.

• Walnut: reads your real connected holdings. Walnut connects your real brokerage (Yes (your own broker)) and works from your actual positions, so its answers reflect what you genuinely own rather than a generic model.
• Fidelity Go: manages a separate account it holds. Fidelity Go does not read the brokerage you already use. It opens and holds a new account, then invests the money you move into it, so its view is limited to what sits inside Fidelity Go.

This is where they diverge most. One works from your real, connected account while the other does not, so if you want an assistant grounded in the exact positions you already hold, that gap is the thing to weigh first. This holdings-aware angle is the one Walnut is built around: it connects the brokerage you already use and reasons from your live positions, read-only by default, with any trades left for you to approve.

There is no universal winner here; the right pick depends on the job you are hiring the tool for. Match the category to your intent rather than chasing a single “best.”

• Choose Walnut if you want talking to your own brokerage with ai. Its AI conversational + thematic baskets + trade, it is priced as free tier, and it fits chat-driven management of your own brokerage. It is built for hands-on investors who want an AI layer on the broker they already have, without moving their money. Keep in mind that it sits on top of your broker rather than being a broker itself, so you connect an existing account.
• Choose Fidelity Go if you want small balances, where it is free. Its AI automates a fidelity flex fund portfolio, it is priced as free under a stated balance, then a flat percentage (verify current), and it fits hands-off automated investing (robo-advisors). It is built for someone with a smaller balance who wants automation at no cost and holds it in a retirement account where harvesting would not apply anyway. Keep in mind that no tax-loss harvesting, which removes the strongest argument for a managed taxable account.

Because they sit in different categories, this is not strictly either-or: some investors use one for talking to your own brokerage with ai and the other for small balances, where it is free, and just watch for overlapping costs.

Cost is easy to misread when two tools charge in different shapes, so compare the model, not just the number. Walnut is priced as free tier, while Fidelity Go is priced as free under a stated balance, then a flat percentage (verify current). A percentage-of-assets fee scales with your balance, a flat subscription does not, and a “free” tier usually earns elsewhere (on cash, order flow, or premium upgrades), so the cheapest headline is not always the cheapest outcome for your situation.

Pricing and tiers change often. Confirm the current numbers on each provider’s own site before you decide; the framing above is point-in-time.