Chaz Englander used to run a marketplace where neighbours rented each other’s drills and cameras. His brother Arnie ran a grocery delivery startup. Both companies were sold, and the one the brothers built together is now worth $1.5 billion.

That company is Model ML, a London business that builds AI agents for bankers, consultants and investors. On November 17, 2026, it announced a $250 million Series B at a $1.5 billion valuation. A Series B is a company’s second big round of venture funding, usually raised once the product is selling and the job is to grow.

General Atlantic and QED Investors led the round together. Evolution Equity Partners, Airtree, Forestay Capital, Y Combinator, FT Partners, Latitude, Inception Fund and 13Books Capital joined in. Total raised so far: $350 million.

Chaz and Arnie, before the banks

Chaz founded Fat Llama, a rental marketplace that went through the Y Combinator startup programme in 2017 and was later bought by Hygglo. Arnie built Fancy, which joined the same programme in 2020 and ended up with Gopuff.

They started Model ML in 2023 and went back to Y Combinator a third time the following winter. Chaz is chief executive.

The target this time was the work junior bankers lose their evenings to. Financial models. Pitch decks. Due diligence reports.

Model ML’s answer is a personal agent for each employee. An AI agent is software that carries out tasks by itself instead of waiting to be asked a question. This one learns a banker’s clients, live deals and preferences, runs on the firm’s own systems around the clock, and hands back analysis in the firm’s house format.

Agents are popular with investors this year, and Desible.ai’s ₹32 crore round for AI phone calls is one recent example. Chaz Englander thinks most AI tools still stop at answers and alerts.

“It should understand your work and get things done.”

Chaz Englander, Co-founder and CEO, Model ML

The decision to stop renting all its intelligence

Most AI startups build on top of a model from one of the big labs and pay for every request. Model ML began that way too. The bigger move, laid out in General Atlantic’s announcement, is that the company now runs a growing share of its work on open-source models it has trained further itself. An open-source model is one whose inner workings are published, so anyone can adapt it. Model ML tests models against real finance tasks and sends each job to whichever does it best on quality, speed and cost, and it says its own versions cost 2% of what the top commercial models charge, usually run five times faster, and match or beat them on accuracy for the tasks they handle. Those are the company’s figures, and they explain a lot about the valuation. The price of running AI is the same problem behind Fractile’s £577 million Series C and Torus selling AI computing from a flywheel business.

Banks also care where their data goes. Model ML often installs its software inside a customer’s own infrastructure, and says customer data is never used to train AI models.

Twelve times the revenue in a year

In the third quarter of 2026, annual recurring revenue was 12 times what it had been a year earlier. That term means the yearly value of the subscriptions customers have signed up for. Daily active users grew 26 times.

The money arrived quickly as well. A $12 million seed round came in February 2025, and a $75 million Series A led by FT Partners followed that November. The Series B lands 18 months after what the company counts as its launch from stealth, the period when a startup works without publicity.

Its advisory board includes Sir Noel Quinn, former group chief executive of HSBC, and Axel Weber, former chairman of UBS.

Hang on, twelve times what?

That was my first question, and the announcement doesn’t answer it. There is no revenue figure in dollars and no customer named. A 12-fold jump from a small base is still a small number.

Two things make me take it seriously anyway. General Atlantic and QED have seen the real figures and put in $250 million. And the 2% cost claim, if it holds up inside a bank, gives Model ML an advantage that comes from its own engineering and can’t simply be bought from an AI lab by the next competitor.

I also like that the founders learned to sell with drill rentals and grocery orders. Banks are a tougher crowd.

Where the $250 million goes

Model ML says it will hire more researchers and engineers in London and New York, keep developing the product, and support larger rollouts at customers. It opened offices in Paris, Madrid, Tokyo, Singapore and Dubai this year to serve them.

Laura is a business historian and researcher who writes about how companies began and grew. At Companies History, she works primarily from original sources such as annual reports, regulatory filings, patents, and archived news coverage. Her articles often correct popular myths about famous founding stories.