How Capital One Built Its Own AI Brain Instead of Buying One Off the Shelf

Key points

• Capital One's multi-agent AI system, named MACAW, handles millions of bank fraud calls a year across interactions lasting four to sixty minutes.
• The bank fine-tunes open-weight models, AI models whose underlying code is publicly available, using proprietary customer data that outside providers cannot access.
• Capital One's Chat Concierge car-shopping assistant runs on a customised version of Meta's Llama, an open-weight large language model.
• Capital One's VP of machine learning engineering Kel Vanee said routing queries across multiple models can beat the accuracy of any single model alone.
• Vanee predicted that proactive, event-driven AI, software that acts without waiting to be asked, will be a major near-term trend in financial services.

Most big banks quietly license an AI model from one of the tech giants and call it done. Capital One took a different path.

At VB Transform 2026, first reported by VentureBeat, Kel Vanee, the bank's VP of machine learning engineering, laid out why Capital One decided to build its own AI infrastructure from the ground up rather than rely on ready-made products.

"At Capital One, we're not just using AI, we're building AI," Vanee said.

What did Capital One actually build?

The centrepiece is MACAW, short for Multi-Agent Capital One Workflow. A multi-agent system is software made up of several separate AI programs, each with a specific job, that pass work between them like a relay team.

Fraud calls were the proving ground. A single large language model, the kind of AI that reads and writes text, could not handle the full complexity of a fraud conversation that might run an hour. So Capital One split the work across four agents.

First, an understanding agent works out what the customer actually wants. A reasoning agent then drafts a summary of the call. A validation agent checks that summary for accuracy. Finally, an explaining agent turns it into a clean document for the human specialist who takes over.

The result: several hundred customer-service agents who handle complex fraud cases no longer have to reconstruct long calls by hand. The system writes the post-call notes for them.

Why build on open-weight models instead of buying a finished product?

Capital One could have licensed a frontier AI model, a polished commercial product from companies like OpenAI or Google. Instead it chose open-weight models, AI whose underlying code is publicly available and can be modified.

The reason is data. Capital One holds decades of transaction records, fraud patterns, and customer interactions that no commercial AI provider can touch. By training open-weight models on that private data, the bank bakes its own expertise directly into the AI.

"We view our data as a huge advantage," Vanee said. "Something that nobody else has."

A useful side effect emerged: tuning a model for one job lifted performance across other tasks too. Training the AI to understand Capital One's internal policies and terminology made it better across the board.

The bank's customer-facing car-shopping assistant, Chat Concierge, runs on a customised version of Meta's Llama, a widely used open-weight model. It uses the same four-agent structure as the fraud workflow.

What comes next for the bank's AI?

Vanee outlined two priorities. The first is model routing, where a system sends each query to whichever AI model is best suited to answer it, rather than sending everything to one model. Vanee believes this approach can deliver better accuracy than any single model at lower cost.

The second is proactive AI. Today most AI waits for a person to ask a question. Vanee wants systems that detect a problem and act before anyone asks. In fraud terms, that means the AI spots suspicious activity and flags it without waiting for a customer to call.

"Proactive AI is going to be a really important trend," he said.

Common questions

Does this affect how Capital One customers experience the bank?

Yes, directly. The MACAW system already handles post-call notes for fraud specialists, meaning agents spend less time on paperwork and more time helping customers with complicated cases.

Is an open-weight model less secure than a commercial one?

Not by default. Security depends on how a model is deployed and governed. Capital One says it built governance and guardrails into the platform itself, so every agent operates within defined rules regardless of which underlying model powers it.