This is the kind of product that sounds empowering and ends up quietly changing who people blame when they lose money.
An AI trading assistant that lives inside a chat — and doesn’t just help you “research” stocks, but can also execute trades right there — is convenient in the same way one-click buying is convenient. It removes the speed bumps. And speed bumps exist for a reason.
Based on what’s been shared publicly, Liquid has launched “Co-Invest” on Grok. The pitch is simple: you can talk to it, ask what’s moving a stock, dig into “catalysts,” and then place trades without leaving the conversation. It’s already pulled in over 10,000 users and run more than 5 million AI tool calls. Whatever you think of those numbers, they tell you one thing: people want this. A lot.
Here’s my judgment: conversational investing is a power tool being handed to people who mostly think they’re being given a seatbelt.
The upside is real. If you’ve ever watched someone bounce between a broker app, a news app, a chart, a search tab, and a notes app — just to answer “why is this stock down today?” — you understand the appeal. Imagine you’re at work, you see a stock spike, you ask the assistant what happened, it summarizes what it can find, and you decide whether to buy or sit tight. You save time. You feel more informed. You don’t need the “traditional” finance background just to understand basic context.
But the danger is baked into the same feature that makes it attractive: one conversation can smoothly slide from information to action. “Here’s what might be happening” turns into “so should I buy?” and then into “okay, do it.” If the assistant is wrong, vague, or overly confident, the trade still happens. If it’s right for the wrong reasons, the user still learns the wrong lesson.
And people will learn lessons from this, fast.
Picture a new investor who doesn’t know how to read filings or spot hype. They ask the assistant why a company is trending. The assistant gives a clean story, because clean stories are what chat tools are good at. The user feels clarity, hits buy, and gets a rush. That rush is not “better investing.” That’s a feedback loop: confidence, action, reward (sometimes), repeat.
Now picture a different person: someone who already trades too much. For them, a chat-based execution flow is basically a turbo button. The friction that used to slow down impulsive trades — logging in, double-checking, second-guessing — shrinks to a couple of messages. If you think overtrading is a problem now, wait until “talking yourself into a trade” is literally the product interface.
The company will say (and maybe believe) this helps people make informed decisions. Maybe it does, sometimes. But “informed” isn’t the same as “wise,” and it’s definitely not the same as “suitable.” An assistant can explain what a market catalyst is and still push a person toward risk they don’t understand. The assistant can lay out pros and cons and still leave the user with one strong emotional impression: urgency.
The most uncomfortable part is accountability. When a human adviser gives bad advice, you at least understand the relationship: they’re a person, they have incentives, they might be wrong. When a chat assistant nudges you into a trade, it feels like “the system” recommended it. If it goes badly, who takes the blame? The user will blame themselves, or the market, or the AI, or the platform. That confusion is not a small detail. Confused accountability is how bad products survive.
There’s also the platform question. If this happens “within conversations,” then the conversation is the funnel. The UI isn’t a calm, boring trading screen anymore. It’s a persuasive space. Even if no one intends it, chat has a way of guiding you. It answers quickly. It fills silence. It gives you something to do next. That’s great for customer support. For trading, that can turn into pressure.
To be fair, there’s a world where this helps. Imagine a cautious investor who mainly buys long-term index funds, but occasionally wants to understand a company they work with. The assistant can translate jargon into plain language, highlight risks, and stop them from making a dumb move based on a rumor. Or imagine someone outside finance who finally feels like they can ask “stupid questions” without being judged. That matters.
But the product isn’t just “ask questions.” It’s “ask questions and trade immediately.” The second part is the part that changes behavior. If you want to help people learn, you build in pauses. You create moments that force reflection. You don’t make execution the easiest step in the whole flow.
The early traction tells me this is going to spread, because it matches how people already behave: they talk first, they decide in the moment, they act before the feeling passes. The winners will be the platforms that capture that moment. The losers will be the users who confuse a smooth conversation with solid judgment.
So here’s what I actually want to know: should an AI that can execute trades inside a chat be designed to slow people down on purpose, even if that hurts growth?