Capital One has 12 dated AI-agent moves on the record, from Mar 4, 2025 to Sep 29, 2026, and 4 agent use cases on its profile, 4 in production (Chat Concierge, Brex agentic payments, VulnHunter and AI and model risk in the 10-K). The latest move, on Sep 29, 2026: Natarajan: cutting latency and token costs with smaller, open-weight models (Fortune). Each row below is the bank's own dated entry with the stage it describes and its source; the strategy, numbers, leaders and regulators are on Capital One's AI strategy page, and the same rows sit in date order with every other bank on the tracker.
What AI agents does Capital One run, and at what stage?
Capital One's profile records 4 agent use cases in production: Chat Concierge, Brex agentic payments, VulnHunter and AI and model risk in the 10-K.
- In production
Chat Concierge
Multi-agent assistant on dealer websites; compares vehicles, schedules appointments, routes to sales.
- In production
Brex agentic payments
Agentic AI products for corporate cards and expenses, acquired with Brex.
- In production
VulnHunter
Agentic code-security tool with attacker-first analysis and a falsification engine; open-sourced in July 2026.
- In production
AI and model risk in the 10-K
The 2025 10-K names models and AI in pricing, fraud detection, loan grading, credit decisions and software development as a risk, plus agentic-commerce fraud and AI intellectual-property uncertainty.
What has Capital One announced about AI agents, and when?
Sep 20263 moves
Natarajan: cutting latency and token costs with smaller, open-weight modelsAnnounced
Chief enterprise AI officer Prem Natarajan says the bank has put more emphasis on agentic AI infrastructure to reduce latency and lower token costs, building customised open-weight and specialised models with less context targeted at the desired output; customer-service agents now edit only one-fifth of AI-drafted copy.
Shetty: 'agentic AI now must be treated as an end-to-end system'In production
VP of enterprise AI Rashmi Shetty says 14 years of data-infrastructure work gives agents a strong foundation and context, and that the bank adopts platform-first governance — rule-based validation, sandbox simulations, observability of agent trajectories, standardised harnesses and human-in-the-loop controls for high-risk actions — before developers build agents, rather than retrofitting it later.
Agentic-commerce principles with five other banksAnnounced
With Bank of America, CBA, ING, NatWest and ASB, a paper on transparency, safety, privacy and data, choice and interoperability.
Aug 20261 move
MACAW and open-weight models, describedAnnounced
A machine-learning engineering leader explains the multi-agent fraud-call workflow and fine-tuning open-weight models on proprietary data.
Jul 20261 move
VulnHunter open-sourcedIn production
An internally built agentic AI code-security tool released on GitHub under Apache 2.0; the CISO cites the brief window before AI attack tools spread.
Apr 20262 moves
ICLR 2026: alignment and guardian-model researchAnnounced
Gold sponsor and contributor on LLM alignment, uncertainty and responsible agentic systems, including the DynaGuard guardian model.
Brex deal closesAnnounced
Brex, which Capital One's January announcement said serves over 25,000 companies, brings a suite of agentic AI products to the business-banking strategy.
Feb 20261 move
2025 10-K: 'our evolving use of AI' is a risk factorIn production
Models and AI in pricing, fraud detection, loan grading, credit and software development; agentic-commerce fraud liability; uncertain IP protection for AI output.
Jan 20261 move
Q4 2025 call: a stack built 'working backwards from the AI revolution'In production
Fairbank says the bank is 'now building AI solutions across our businesses' and that Brex has deployed in-house AI agents for expense management and audit.
Nov 20252 moves
Applied AI research agenda publishedAnnounced
LLM training, customisation and alignment; synthetic data; multi-agent reasoning.
NeurIPS 2025: 20-plus accepted papersAnnounced
Responsible AI, multi-agent systems, LLM evaluation and fraud detection.
Mar 20251 move
Chat Concierge launches for auto dealersIn production
A multi-agent assistant that compares vehicles, books appointments and hands off to sales, on the Auto Navigator platform.
What has been reported about Capital One's agents?
Headlines from the Capital One news stream, publisher and date only:
- Capital One’s agentic AI strategy hinges on data, platform-first mindset
ciodive.com · Sep 28, 2026
- Why Capital One built its multi-agent AI platform around open-weight models
venturebeat.com · Aug 13, 2026
What AI agents does Capital One have in production?
Capital One's profile records 4 agent use cases in production: Chat Concierge, Brex agentic payments, VulnHunter and AI and model risk in the 10-K.
What has Capital One announced about AI agents, and when?
12 dated moves from Mar 4, 2025 to Sep 29, 2026: 5 describe an agent in production, 0 a rollout, 0 a pilot and 7 an announcement. The latest, on Sep 29, 2026: Natarajan: cutting latency and token costs with smaller, open-weight models (Fortune).
Where does this timeline come from?
From Capital One's profile on this site, which dates each move and cites its source; the build checks every link. The stage is the one the bank describes, and nothing here is written by a model. The profile is refreshed on a biweekly routine; the record last changed on Oct 19, 2026.
Other banks with an agent timeline:
When Capital One puts an agent into production, you'll read it the next morning.
when one of the banks we cover moves on AI, it is in the next morning's brief · six sourced stories · 7 am PT · free
plus every regulation, bank and agent page update, the morning after · leave any morning