Kraken Launches an AI Assistant That Suggests Trades on Its Own

Crypto exchange Kraken is rebuilding its mobile app around an AI system that watches the market on its own and suggests trades to the user. The company calls it "financial intelligence." It studies a person's portfolio and financial goals, then recommends specific steps. The buy or sell button still gets pressed by the person, not the machine. Similar tools have already shipped from OKX, Coinbase, and Revolut, so Kraken is joining a wave that has spread across nearly every major exchange in recent months.

What exactly is Kraken changing?

According to the company's own announcement, the new experience starts with a simple step. Users set a financial goal first, such as buying a home, saving for retirement, or building an emergency fund. The app then reshapes its interface and recommendations around that goal instead of forcing people to figure out complex trading tools on their own.

A typical crypto app used to greet users with charts, an order book, and dozens of settings on the very first screen. Kraken now removes that barrier and builds the whole flow around plain questions. How much money is needed. By when. For what exact purpose. Terms like limit orders or margin fade into the background.

How does Kraken's financial intelligence actually work?

After a user picks a goal, the system asks about risk tolerance, preferred funding methods, and a broader financial profile. Based on that data, it builds a portfolio the person can review and adjust before putting any money in. From there, the assistant keeps tracking positions and sends tailored updates after every meaningful market move.

• Core idea: the AI constantly scans the market for trades that fit a specific user's stated goals.
• Every recommendation requires manual approval; there is no automatic execution built into the system.
• The portfolio can be rebuilt around a new goal or a changed risk level at any time.
• Once invested, the assistant sends short prompts instead of a full technical report.
• Unlike classic robo-advisors, which rebalance a portfolio once a quarter, this reacts to the market almost right away.

Why are exchanges rushing to build agentic assistants?

Kraken is far from the first mover here. In June, OKX launched a beta marketplace where AI agents can transact autonomously, complete onchain tasks, and build their own reputation on the blockchain. That same month, Coinbase rolled out a tool letting agents pay and trade crypto on a user's behalf through its x402 payments protocol.

Chainalysis data shows agentic payment activity on Coinbase's Base network has crossed 100 million transactions. Growth in the raw number of transfers has leveled off, but the share of larger transfers has climbed noticeably. That points to AI-driven payments moving past small micropayments and early testing. On Friday, fintech firm Revolut upgraded its own Revolut X exchange. Customers can now connect assistants such as Claude, Gemini, or Cursor to analyze markets and place orders in plain language, also with mandatory approval before each trade.

Robinhood shows a similar pattern. According to the company, since it launched a beta version of its AI agent feature in late May, equities and options traders have already opened more than 70,000 accounts with the feature enabled. That points to demand coming not just from a narrow circle of tech enthusiasts but from everyday app users too.

How safe is it to let AI weigh in on trading decisions?

Kraken insists this is a decision-support tool, not an automated trading system. The company's chief data officer explained the reasoning behind the launch.

"There's an opportunity for everyday people to become high-frequency traders and do so using plain English."

- Kamo Asatryan, Chief Data Officer at Kraken, in an interview with CNBC

The gap with OKX's approach is real. Those agents act on their own and build a reputation on-chain without a human at every step, while Kraken and Revolut always leave the final call to the user. The risk does not disappear either way. A model can produce a weak recommendation, and a user who trusts the tech too much may accept it without checking the numbers. For now, none of the major exchanges let an agent trade fully without the account owner's sign-off. A lot also depends on the quality of the data behind the model. A weak signal from the algorithm can look just as convincing as a strong one.

What does this mean for Bitcoin and Ethereum traders?

For holders of Bitcoin and Ethereum, the practical effect is simple. Routine market analysis gets handed to an algorithm, while the final call on any trade stays with the person. That lowers the entry barrier for newcomers who used to avoid complex terminals, while giving active traders one more tool for price monitoring.

Competition between exchanges for user attention is now built around interface convenience, not just fee size. Expect similar features to show up in other major apps in the coming months. The market for agentic trading assistants is only starting to take shape. The bigger question for the next phase is not whether every exchange gets an AI layer, but how good its advice turns out to be once the market suddenly turns.