Model ML raises $250 million to put AI agents inside banking workflows
General Atlantic and QED Investors co-led the round. The startup is pitching agents that prepare finished work without waiting for instructions, with performance figures that remain company claims.
Model ML says its agents can monitor company news, judge its relevance to live deals, and produce or update financial models and presentations in minutes. The London startup reported third-quarter year-over-year growth of 12× in annual recurring revenue and 26× in daily active users, but disclosed neither absolute revenue nor user counts. Its $250 million Series B values it at $1.5 billion; the company plans to use the funding to grow research, engineering, and customer deployment capacity.
01
The agents integrate with customers’ systems, including Excel and PowerPoint, and are intended to handle financial models, pitch decks, and due-diligence reports.
02
Model ML says its fine-tuned open-source models cost 2% as much as frontier models and are usually five times faster on assigned tasks; it claims comparable or better accuracy on those tasks.
03
The company says it can deploy inside customer infrastructure and does not use customer data to train AI models; it lists ISO/IEC 27001:2022 certification and a SOC 2 Type II audit.
Model ML is promising bankers something beyond a chatbot: an AI agent that notices relevant developments and prepares finished analysis before anyone asks. Investors have put fresh money behind that pitch. On November 17, 2026, the company announced a $250 million Series B at a $1.5 billion valuation, co-led by General Atlantic and QED Investors.
The financing brings Model ML’s total funding to $350 million. It follows a $75 million Series A led by FT Partners in November 2025, placing another, substantially larger round behind the business a year later. General Atlantic’s announcement says the company emerged from stealth just 18 months ago.
Other participants included Evolution Equity Partners, Airtree, Forestay Capital, Y Combinator, FT Partners, Latitude, Inception Fund and 13Books Capital. Brothers Chaz and Arnie Englander founded the London-headquartered company, whose intended users span banking, consulting and investment work rather than a single financial specialty.
Alongside the financing, Model ML disclosed sharp increases in recurring revenue and daily usage in the third quarter. Those figures describe growth, not the size of the business: the announcement gives neither an absolute annual recurring revenue figure nor a daily active user count.
Model ML’s reported third-quarter growth
12×Annual recurring revenue
Model ML says annual recurring revenue grew 12 times year over year in the third quarter of 2026.
26×Daily active users
The company says daily active users increased 26 times in the third quarter of 2026.
From company news to a finished presentation
The work Model ML targets is concrete: building financial models, preparing pitch decks and writing due-diligence reports, according to TechCrunch. These are tasks traditionally performed by junior bankers and consultants. Its agents connect with clients’ systems, data and tools, including Excel and PowerPoint, rather than operating only as a separate place to ask questions.
Model ML’s more ambitious claim is that the software can initiate that work. Its example starts with news about a company a banker covers. The agent assesses whether the development matters, works through its implications and prepares analysis in the firm’s preferred format. Model ML says this takes minutes rather than days.
The startup also says its agents understand the deals users are working on and can automatically update financial models and presentations when developments change the analysis. That pitch depends on context: Model ML describes agents that know a professional’s clients, live deals and working preferences, becoming more personalized through interactions.
Access to that information comes with deployment choices. Model ML says it often runs within customers’ own infrastructure, so their data need not leave their environment. It also says customer data is never used to train AI models. The company lists ISO/IEC 27001:2022 certification and a SOC 2 Type II audit.
A task-specific case for cheaper models
The model strategy is not exclusive reliance on one AI supplier. Model ML says it benchmarks models on financial work, then routes tasks according to quality, speed and cost. A growing share goes to its own fine-tuned open-source models: existing models it has further trained, rather than a single outside provider handling every request.
For the workloads those in-house models handle, the company claims they run at 2% of the cost of frontier models and are, in most cases, five times faster. It also claims accuracy matches or beats those leading models on the same tasks. These are Model ML’s own comparisons, not a claim of equivalent performance across all financial work.
General Atlantic’s financial-services head, Paul Stamas, framed the investment around institutions moving from AI experiments to everyday use. He said customers were trusting Model ML with increasingly complex work. That is the investing firm’s assessment of adoption, alongside the company’s own growth figures, rather than a separate measure of the agents’ accuracy.
Model ML says the new capital will expand its research and engineering teams in London and New York to develop the product and support wider customer deployments. Offices opened this year in Paris, Madrid, Tokyo, Singapore and Dubai will support service for clients across those markets.
Sources
generalatlantic.comModel ML raises $250 million at $1.5 billion valuation, 18 months after launching from stealth | General Atlantic
techcrunch.comModel ML raises $250M to automate investment banking grunt work | TechCrunch
Reader comments
Newest comments first. Replies stay oldest first.