What is agentic trading?
Agentic trading is the next step past automated trading: instead of a system that only follows pre-set rules, an AI agent makes and executes investment decisions on your behalf, continuously, based on goals you set once rather than instructions you give trade by trade. Not a standing instruction that only fires when a price is hit. An agent that ingests price feeds, filings, and news in real time, rebalances a portfolio, and acts, often faster than any human could read the same information.
The closest human equivalent is a discretionary wealth manager, someone who makes investment calls for you rather than just carrying out orders you give them. The difference is speed: a human advisor reviews your portfolio quarterly, or monthly if you’re paying for it. An agent reviews it continuously and can react to good or bad company news within seconds, not days.
What is agentic commerce?
Agentic commerce is an AI agent completing purchases on your behalf: searching, comparing, and checking out, across merchants, without you opening a single tab. You set the budget, the size, the delivery window, and the agent optimises for it.
The closest human equivalent is a personal shopper who only knows a handful of stores well, while an agent checks practically any merchant selling online, all at once. The agent still can’t replace a shopper’s taste, but it can handle the comparison shopping behind purchases.
Agentic trading works the same way, but in reverse: it swaps a wealth manager’s judgment for a system’s, one that earns trust through transparency and a track record rather than fiduciary duty.
Why agentic trading is a big shift
Institutional trading desks have run algorithmic and quant strategies for over two decades. What’s new is that reasoning-capable agents push that same real-time, information-driven decision-making down to individual retail accounts, which until now got sorted into one of a handful of standard portfolios and left largely untouched between check-ins.
The asymmetry this closes is access, not existence. A hedge fund already had a system that could read an earnings call transcript and reposition in seconds, but a retail investor never has, until now. Agentic trading means that capability is no longer gated behind a six-figure account minimum.
Because the decisions happen at machine speed, so does the risk of getting one wrong: a flawed model will make the same bad call repeatedly until something catches it. For this reason, higher returns don’t necessarily make a better agentic trading product if it doesn’t have sufficient guardrails. These products need position limits, explainable reasoning traces, and a human circuit breaker that can pull the plug.
What it means for individual investors
Agentic trading removes the two things that have historically kept ordinary people out of managed investing: the account minimum and the knowledge requirement. A discretionary wealth manager typically won’t take you on below a six-figure minimum, and charges roughly 1% of assets annually for the privilege. An agent has no minimum, and it doesn’t require you to already understand diversification, rebalancing, or how to read a prospectus before it can act on your behalf. Tell it your goal and how much risk you’re comfortable with, in plain language, and it will build a portfolio around that.
The part worth being clear-eyed about is who the agent actually works for. A human fiduciary advisor, someone legally bound to act in your interest, has that obligation regardless of who pays their salary. An agent’s incentives are only as good as whoever built and deployed it, and unless the provider is explicit about acting in your interest the same way, an agent optimised for engagement or for steering you toward in-house products isn’t operating under the same obligation, even if it looks and talks like an advisor.
The other real gap is accountability when something goes wrong. If a human advisor gives you negligent advice, there’s a regulatory and legal path to recourse. That path is much less defined for an AI agent acting autonomously, and it will take years of regulation catching up before “the agent made a bad call” has a clear answer for who’s liable.
Why agentic commerce is a big shift
Once an agent is doing your shopping, your side of it gets almost boring. You describe what you want, a budget, a size, a delivery window, and it buys it. No five tabs open comparing prices, no wondering if you’re actually getting a good deal, no falling for a countdown timer at midnight. That simplicity is possible because the tactics that have shaped online shopping for two decades don’t work on an agent the way they work on you.
It’s not a small shift, either. McKinsey’s $3 trillion to $5 trillion global forecast for 2030 reflects a meaningful share of existing retail spend moving through agents instead of storefronts and search results, not e-commerce simply growing faster. Shopping itself is being redefined here, the boundaries between platforms and services collapsing into a single intent-driven flow: you describe what you want, and the agent handles the rest.
Three specific tricks stop working the moment your agent is the one deciding:
- Urgency pricing loses its grip. Flash sales and countdown timers work on a human because watching a clock run down triggers loss aversion. Your agent doesn’t feel anything watching a clock, it checks the offer against its actual terms and walks if a better one exists elsewhere.
- Paid placement stops paying off. Search ads and sponsored listings work because a human is scrolling and a retailer paid to interrupt that scroll. Your agent isn’t scrolling, it’s querying, and it has no reason to rank a paid result above an unpaid one that better matches your criteria.
- Price discrimination runs out of room to hide. Dynamic pricing and retailer-specific markups survive because comparison shopping across five sites costs a human real time and effort. It costs your agent nothing, so every purchase becomes a five-tab comparison, every time, the one thing your agent is built to never skip.
These three tactics fund a large share of how retail makes money today. Global digital ad spend alone runs past $700 billion a year, much of it paid specifically for the kind of placement an agent has no reason to honour.
What it means for retail once agents are the customer
Short term, agents doing the comparison shopping will mean better prices and less time spent. The harder question is what you’ll actually notice once retailers adapt, and it splits three ways.
- The toll booth just moves. Search ads and sponsored listings used to sit between you and a fair comparison. Once your agent queries merchants directly, that toll booth moves earlier, to API access, verified merchant status, and priority in structured product feeds, so you may still end up seeing options shaped by who paid for access, just one step further removed from where you can see it.
- A level playing field, but only for commoditised goods. For anything fully specified, a flight, a generic electronics item, price and fit are the only inputs that matter, and your agent will find you the best one regardless of how big or small the retailer behind it is. For anything where trust matters more, quality control, counterfeit risk, your agent defaults to merchants it already trusts, the same instinct you’d have shopping it yourself.
- Scale wins anyway. If this turns into a pure price war, you’ll keep ending up with whoever has the strongest logistics and lowest per-unit cost, usually the incumbent, not the local business, since an agent optimising purely for price has no loyalty to either.
What that means for you: the retailers you actually end up buying from won’t necessarily be today’s biggest names. They’ll be whoever has invested in being legible to an agent, structured data, clean APIs, transparent fulfilment, whether or not you’d ever heard of them before.
The technical differences
Data environment
Trading agents work with dense, structured data: price feeds, order books, filings, macro indicators. Commerce agents work with messier, unstructured data: product listings, reviews, shipping times, return policies, scraped from sites never built to be read by machines, which takes far stronger tools to extract and normalise.
Latency requirements
Trading agents often need to act in milliseconds, since a delayed decision can mean a worse price or a missed window entirely. Commerce agents run on a far more forgiving timescale: thirty seconds spent comparing five retailers costs nothing close to what a delay costs in trading.
Reversibility
A bad trade can usually be undone at a cost, since you can sell a position back into the market. A bad purchase is often harder to reverse, shipping fees, restocking fees, a gift that already arrived late, so commerce agents need to weigh returns and cancellation windows more conservatively.
Regulatory environment
Trading agents operate inside a dense regulatory framework covering identity verification, suitability, and market manipulation. Commerce agents operate in a much lighter one today, though that’s shifting as agentic checkout becomes common enough for consumer protection law to catch up.
Payment mechanics
Trading agents move money within regulated brokerage accounts built for exactly this. Commerce agents have to authenticate and pay across dozens of independent merchants, each with its own checkout flow, which is why multi-currency, agent-friendly payment infrastructure matters so much for agentic commerce to work at scale.
The common thread
Both shifts hand a decision to something that sees more and acts faster than you can, but the downside differs sharply: a bad trade can erase a meaningful chunk of a portfolio before you notice, while a bad purchase costs a restocking fee. That gap is why agentic trading still needs guardrails before it goes mainstream, while agentic commerce is already undercutting the pricing and advertising models retailers built their margins on, one comparison-shopped purchase at a time.
Get ahead of it
Liminal already builds in the human circuit breaker this piece describes. Nim, Liminal’s AI agent, places trades for you, tracks your portfolio, explains why a stock moved, tells you whether it still matches the balance you set, and flags what relevant news could mean for your holdings. Nim won’t offer advice, just information, and Nim never places a trade without your sign-off first.
That makes it the bridge between where consumer fintech is today and full agentic trading. You get most of the access that agentic trading promises, with no minimums, no need for existing investing experience or financial knowledge, and without the “who does the agent answer to” risk, since you still confirm every trade.