Kraken, one of the largest US crypto exchanges, wants to put artificial intelligence at the center of how people trade on their phones. The company says it is rebuilding its mobile app around "agentic trading," in which AI agents research opportunities, offer advice and can execute trades on a user's behalf within preset limits, The Block reported. One caveat matters up front: this is a described plan, not a shipped product, and Kraken has not given a firm launch date.
What "agentic" trading means
The buzzword is worth unpacking. A chatbot answers questions; an AI "agent" takes actions. In Kraken's telling, the agent would learn a user's goals and risk tolerance through plain-language prompts, watch markets continuously, surface relevant news, and, with the user's authorization, place trades and suggest moves such as putting idle cash to work. The pitch is to make sophisticated, always-on trading accessible to ordinary users who would describe what they want in everyday language.
Part of a bigger expansion
The app overhaul fits a broader strategy to turn Kraken from a pure crypto venue into a multi-asset broker. The company has moved into tokenized versions of US stocks and round-the-clock trading, and, according to CNBC, has confidentially filed for an initial public offering, a step toward listing its own shares. Adding AI-driven tools is a way to attract more active traders and stand out as crypto exchanges increasingly compete with mainstream brokerages.
The risks of letting software trade
Handing trading decisions to an AI, even a supervised one, introduces hazards that regulators and courts have barely begun to work through.
The first is that errors can compound fast. Unlike a person who might pause, an agent can act continuously, so a flawed model or bad instruction could trigger a cascade of trades before a user notices. The second is liability: if an AI-placed trade goes wrong, it is not yet clear who is responsible, the platform, the user who switched the agent on, or no one in particular. The third is suitability. Human advisers are generally required to make sure their recommendations fit a client's circumstances; an automated agent running across many users has no such judgment unless it is carefully constrained. Leverage and volatile crypto markets can magnify all of these.
None of that makes the idea unworkable, but it puts a premium on guardrails, spending limits, clear disclosures, audit trails and a way for humans to intervene, being in place before, not after, such tools reach large numbers of people.
Why it matters
Kraken's move is a marker of where consumer finance may be heading, as AI shifts from answering questions to carrying out tasks with real money at stake. If it works, agentic trading could reshape how retail investors interact with markets. But it remains a bet on an unproven product in a lightly settled area of regulation, and the sober reading is to watch how it is built, and how it is overseen, before judging whether it is a genuine advance or simply a faster way to make mistakes. This article is informational and not investment advice.
