Cleo adds voice and memory to its AI money coach, targets $500m ARR

Cleo, the London-based money app, has launched Cleo 3.0, adding voice conversations, long-term memory and a more powerful reasoning engine to the assistant it has been building since 2016. The update, announced on 29 July, is the clearest sign yet that the company wants to be judged as a financial coach rather than a budgeting chatbot.

The app now lets users speak to Cleo directly and get spoken answers back, instead of typing into a chat window. It also remembers goals and habits across sessions, so a user does not have to re-explain their situation every time they open the app. A new reasoning layer, built on OpenAI's o3 model and offered as an upgrade, works through a user's income, spending and upcoming bills step by step before it answers a question such as whether they can afford a weekend out.

Why memory and voice matter here

Most budgeting apps categorise transactions and show a dashboard. Cleo has instead built its reputation on a conversational interface with personality, and the 3.0 release pushes that further by making the assistant persistent rather than stateless. Founder and chief executive Barney Hussey-Yeo framed the shift as closing a gap between mass-market apps and the kind of advice only wealthy clients could previously buy.

"Cleo remembers your goals, learns your habits, and delivers personalised financial guidance previously only available to the wealthy," Hussey-Yeo said.

He also described the release as more than a routine feature drop, saying it signals a shift in what people should expect from financial technology and arguing that the next phase of the AI race will be won by whoever applies intelligence to hard, human problems rather than whoever builds the biggest model.

The company says the chain-of-thought reasoning in the new model lets Cleo break a complex money decision into a short, specific answer. One example: whether a purchase still fits a budget once rent and bills are accounted for. Cleo also claims user engagement on the app now runs at 20 times the level of a typical banking app. It has not published the methodology behind that comparison.

The numbers behind the relaunch

Cleo's product push comes alongside an aggressive growth target. The company expects to pass 1 million paid subscribers during 2025 and says annual recurring revenue should reach $250m, an increase of 82% on the year before. Hussey-Yeo has gone further in public remarks, saying on LinkedIn that $500m in ARR is "around the corner" and raising the prospect of a stock market listing, asking whether it should be in London or New York.

Those figures, if they hold, would put Cleo among the larger consumer fintech businesses to come out of the UK this cycle. The model is subscription revenue, not lending or interchange fees. The company was founded by Hussey-Yeo and Aleksandra Woźniak out of Entrepreneur First and has grown to roughly 500 staff.

What it signals for European AI fintech

Cleo's bet is that consumer trust in an AI assistant depends less on raw model size and more on memory and natural speech. That bar is rising fast. Foundation models from OpenAI and others are becoming table stakes rather than a differentiator for apps built on top of them. For European fintech founders chasing the same consumer attention, Cleo's subscriber and revenue targets set a reference point. Rivals using similar large language models will face pressure to show retention gains, not just matching features.

The open question is whether the reasoning and memory upgrades hold retention over a full year, rather than produce a launch-week bump. It is also unclear whether a freemium chat product can sustain subscriber growth at the pace Cleo is promising ahead of any listing decision.