Summary
- Model ML's $250m Series B was co-led by General Atlantic and QED Investors, valuing the company at $1.5bn.
- Its agents use financial institutions' own systems and data to support research, analysis and document preparation.
- The company reports twelvefold growth in annual recurring revenue and a 26-fold increase in daily active users, without disclosing absolute figures.
London-based Model ML has raised $250 million to expand artificial intelligence agents designed to perform financial analysis, investment research and other professional services work using organisations’ existing applications and proprietary data.
The Series B round values the company at $1.5 billion and was co-led by General Atlantic and QED Investors. Participants also include Evolution Equity Partners, Airtree, Forestay Capital, Y Combinator, FT Partners, Latitude, Inception Fund and 13Books Capital.
Announced on 17 November, the financing brings Model ML’s total reported funding to $350 million, following a $75 million Series A completed in November 2025. The new capital will support engineering and research teams in London and New York, product development and larger customer deployments.
The company reports substantial growth in demand, with annual recurring revenue increasing twelvefold and daily active users rising 26-fold during the third quarter of 2026 compared with the previous year. It has not disclosed the underlying revenue, customer or user totals.
Agents work within financial systems
Model ML develops software intended to perform sequences of work on behalf of employees in banking, investment management and consulting.
The platform aims to retrieve information from established systems, assess its relevance and prepare outputs using an organisation’s existing documents and working methods.
Financial analysis frequently involves information distributed across research databases, internal records, spreadsheets, presentations and communications with clients.
Employees may need to combine those sources to prepare a financial model, investigate a transaction or assess developments affecting a particular company.
Conventional AI assistants can produce text or analyse supplied material, but they may require employees to gather information and provide detailed instructions before a task can be completed.
Model ML is seeking to reduce that manual coordination by developing agents that can identify relevant work and carry out multiple stages of an assignment.
For example, when new information emerges about a business followed by an investment banker, an agent could examine the development, retrieve supporting material and prepare analysis using the bank’s established formats.
The company describes its agents as increasingly personalised, using information about a professional’s clients, assignments and working preferences to determine how tasks should be handled.
That approach could make AI more useful in specialised environments, provided the system can distinguish relevant information from material that is incomplete, outdated or unsuitable for the intended analysis.
Financial work also requires careful treatment of assumptions and calculations. A document may appear professionally prepared while containing figures that cannot be reconciled with the underlying information.
Organisations therefore need mechanisms to examine generated work, identify its sources and determine whether the conclusions are sufficiently supported.
Proprietary information shapes deployment
Financial services businesses hold information subject to contractual, commercial and regulatory restrictions, including transaction documents, investment positions and confidential client records.
Model ML says its platform supports different deployment arrangements and can operate within customers’ own infrastructure, allowing proprietary information to remain inside their established environments.
The company also states that customer data is not used to train its AI models. It reports ISO/IEC 27001:2022 certification and a SOC 2 Type II audit.
These arrangements may assist institutions evaluating information security, although they do not establish that every deployment automatically meets the requirements of a particular regulator or customer contract.
Financial organisations must still examine identity management, access permissions, information retention and the way individual applications interact with sensitive records.
Agent permissions are particularly important where software can carry out actions rather than merely retrieve information.
Preparing a report presents different risks from editing a financial record, transmitting confidential material or initiating an operation affecting a client account.
Businesses may consequently restrict certain activities or require employee approval before an agent completes actions with material financial or regulatory consequences.
Monitoring and audit records are also necessary to establish what information a system accessed, which actions it performed and whether those activities were authorised.
Specialist models target lower operating costs
Model ML does not depend exclusively on one AI model provider. The company says it evaluates different models against financial tasks and routes work according to quality, processing speed and cost.
It is also developing fine-tuned open-source models for particular financial workflows.
According to the company, those specialised models complete the tasks for which they are designed at approximately 2% of the cost of leading frontier models and, in many cases, five times the speed, while maintaining comparable accuracy.
The figures are company-reported comparisons covering selected workloads. They do not establish equivalent performance across every financial application, nor do they represent independently verified industry-wide benchmarks.
Specialisation can create efficiency advantages where a system repeatedly performs familiar tasks with relatively consistent inputs and expected outputs.
However, financial analysis can also involve unusual transactions, changing market conditions and questions requiring extensive judgement. A model optimised for a narrow activity may not perform equally well when required to handle unfamiliar circumstances.
Routing different tasks between models could allow organisations to use more computationally expensive systems only where their additional capabilities are justified.
The actual financial benefit will depend on the cost of operating the complete service, including application integration, monitoring, employee review and corrections to unsuccessful outputs.
Lower processing charges may offer limited savings if a system regularly requires multiple attempts or substantial human intervention before completing useful work.
Assessment should therefore consider completed workflows and the quality of their results, rather than model prices alone.
International expansion follows rapid reported growth
Model ML was founded by brothers Chaz and Arnie Englander and emerged from stealth approximately 18 months before its latest financing.
Alongside London and New York, the company has opened offices in Paris, Madrid, Tokyo, Singapore and Dubai as it develops relationships with financial institutions operating across international markets.
Its reported twelvefold increase in annual recurring revenue and 26-fold rise in daily active users indicate substantial relative growth, although the absence of starting values prevents an independent assessment of the company’s current commercial scale.
Annual recurring revenue measures contracted recurring business rather than the total economic value produced through a technology platform.
Daily active users provide another indicator of adoption but do not establish how often employees complete valuable assignments or what proportion of their work continues to require manual effort.
Financial services customers may also need time to integrate agents into established operating procedures, especially where the software interacts with systems used for regulated activities.
Successful adoption therefore depends on more than the number of employees accessing the platform. Organisations must determine whether agents improve accuracy, reduce avoidable work and maintain the controls required for consequential financial decisions.
The expansion is taking place as enterprise AI providers increasingly compete to automate complete business processes rather than isolated document preparation or information retrieval tasks.
For established financial institutions, the commercial case will involve comparing the cost of software and implementation against measurable improvements in productivity, quality and service capacity.
Employees may complete some assignments more quickly, but the benefits can be offset if automated work requires extensive checking or introduces additional governance responsibilities.
Model ML’s investment gives it resources to expand its engineering capacity and support customers across several financial centres.
The company’s longer-term position will depend on whether its agents can produce reliable and auditable results within complex financial environments, while delivering savings that remain meaningful after integration, supervision and operating costs are included.
Further disclosure of customer retention, absolute commercial performance and independently measured workflow outcomes would provide a clearer indication of whether its reported growth is translating into sustained enterprise adoption.












