Here’s the strange part of Friday. The U.S. economy lost 23,000 jobs in July — a number nobody was expecting, and normally the kind of headline that sends stocks lower. Instead, the S&P 500 closed at a record high. The reason is almost entirely about the Federal Reserve, not the jobs themselves.

Unlike most of the past few years, the Fed hasn’t been debating whether to cut rates. Heading into this month, several officials were openly discussing a hike, with inflation still running around 3.3% — well above the Fed’s 2% target. The committee’s last vote to hold rates steady wasn’t even close to unanimous: it passed 9–3, with three members pushing for an increase right then. So when the jobs report came in soft, the market didn’t read it as “recession warning.” It read it as “the Fed now has less reason to hike,” and rallied on that alone.

That’s the setup. What I actually want to dig into is what happened next, mostly invisibly, inside the AI-driven investing tools a lot of you are using — Betterment, Wealthfront, Schwab Intelligent Portfolios, Fidelity Go, and the growing list of “agentic” platforms marketing themselves as more than glorified auto-rebalancers.

01 — The old modelRebalancing on a calendar, not on the news

For most of the robo-advisor era, “automated” meant something fairly narrow: build a diversified ETF portfolio based on a risk questionnaire, then rebalance it back to target either on a schedule (often quarterly) or when an asset class drifts more than some threshold — say, 5 percentage points — from its target weight. Tax-loss harvesting ran on similar logic, scanning for losses to realize periodically rather than reacting to a specific news event.

That’s not a criticism — calendar-based rebalancing works, and it’s a big part of why robo-advisors became a legitimate low-cost alternative to human advisors, now managing well over a trillion dollars in U.S. assets. But it’s also, by design, indifferent to a week like this one. A quarterly rebalancer doesn’t care that hike odds swung from 55% to 33% in a single trading day. It’ll get to your portfolio eventually.

02 — What’s different nowEvent-driven portfolios, in theory

The newer generation of platforms is explicitly trying to close that gap. Instead of purely calendar-based checks, some now run continuous or event-triggered exposure analysis — a Fed statement, a jobs report, a sharp single-day move in a sector — and re-evaluate your specific holdings against it, rather than waiting for the next scheduled rebalance window.

Rebalancing logic, side by side

03 — The actual exposureWhat a rate hike does to a portfolio, concretely

It’s worth being specific about what’s at stake here, because “rate sensitivity” is one of those phrases that gets thrown around without being pinned down. The mechanism is duration risk: bonds with longer maturities lose more value than short-term bonds when rates rise, because their fixed payments look less attractive against newly issued, higher-yielding debt. A 25-basis-point hike is small on its own, but it compounds with how it changes rate expectations further out the curve.

None of that requires AI to explain — a decent financial planner would walk you through the same thing. What a well-built AI tool should do differently is tell you, using your actual account, roughly how much of your bond sleeve sits in each bucket, without you digging through a fund fact sheet to figure it out yourself.

04 — Tax-loss harvesting this weekVolatility isn’t only risk — it’s also opportunity

Thursday’s selloff, driven by Middle East tensions pushing oil higher, followed by Friday’s rally on the jobs data, created exactly the kind of dip-then-recovery pattern that tends to trigger automated tax-loss harvesting. If a fund in your taxable account dipped Thursday and your platform sold it at a loss while buying a similar (but not “substantially identical,” per IRS wash-sale rules) replacement fund, that’s the system working as intended — realizing a loss to offset gains elsewhere, while keeping you invested in roughly the same market exposure.

If you saw unexpected trades in your account this week, this is the most likely explanation, not a glitch. It’s worth checking your platform’s activity log rather than assuming something went wrong.

05 — Try it yourselfThe one question worth asking your AI advisor this week

If your platform has a chat interface, scenario tool, or “ask a question” feature, this is a good week to actually use it for something concrete instead of a generic check-in. Ask it directly:

A genuinely useful AI advisor should come back with something specific — which funds, roughly what magnitude, maybe a suggested (not automatic) adjustment. If what you get back is a generic disclaimer about “markets can go up or down,” that’s a useful data point too: it tells you the “AI” in your AI advisor is doing less than the marketing suggests.

06 — Between now and Sept. 16What’s actually worth watching

• 1
  The July CPI report, August 12This is the next real catalyst. A hot inflation print pushes hike odds back up fast; a soft one cements the “Fed holds” narrative. More important than any single day’s jobs headline.
• 2
  Your bond sleeve’s duration mixNot to trade around it — just to know, roughly, how exposed you are before the September decision, so nothing that happens is a surprise.
• 3
  What your platform is actually paying on idle cashWith the Fed funds rate still at 3.5%–3.75%, compare your robo-advisor’s cash sweep yield against a plain high-yield savings account. They don’t always match.
• 4
  Your own reaction, not just the algorithm’sThe tool can rebalance instantly. Whether you also chase the headlines and make manual changes on top of it is a separate, and often more damaging, variable.

The honest takeSome of this is real. Some of it is a label.

To be direct about it: no AI tool can tell you whether the Fed hikes or holds on September 16 — nobody has that information yet, including the Fed itself. What a good tool can do is narrower and more useful than prediction: show you, using your real account, how exposed you are to each outcome, and act on that faster than a human checking in once a quarter would.

The honest gap in the market right now is that “AI-powered” has become a label almost every platform slaps on, regardless of what’s actually running underneath. Some genuinely react to events like this week’s in near real time. Others are still running the same quarterly-rebalance logic from five years ago with a chat window added on top. The prompt above is a decent, free way to find out which one you’re paying for.