Nita Menon

At Capital One, design research helps shape our builds. In this Q&A, Jen Cardello, our Head of Research and Design Operations, shares her ideas on the importance of "learning velocity" and creating experiences that are useful and usable. Check it out https://bit.ly/4jqTyxW

2

#Digitalinclusionweek serves as a reminder that everyone deserves to feel confident, capable and in control of their money. In pursuit of this goal, check out how Capital One partnered this year with @EveryoneOn to hold digital skills trainings in Dallas and Houston, where participants learned to navigate the digital landscape and achieve greater financial empowerment. https://bit.ly/4mPyBy0

1

Who is a woman who’s made an impact on your career? #IWD26 Lynn Reedy far & above anybody else, regardless of gender. Execution/Intensity/Outcome driven. One of a kind and influenced an entire generation at eBay. To this day, my best teacher of all time and carried her timeless lessons for ever after, never deviated. Walk the talk is far from obvious for larger companies, yet she did rely on those who did, regardless of hierarchies, to scale eBay at unbelievable velocity. It showed in growth.

6

Our own EVP, Chief Scientist & Head of Enterprise AI Prem Natarajan recently spoke to TechTarget about Capital One’s partnership with the National Science Foundation (NSF), which will help launch five new National AI Research Institutes to drive breakthroughs in high-impact areas across society like material and computational sciences, economics, and mental health. https://bit.ly/46fLI5I

1

Excited to share that Capital One has entered into an agreement to acquire Brex, a modern, AI-native software platform offering intelligent finance solutions that make it easy for businesses to issue corporate cards, automate expense management and make secure, real-time payments. https://bit.ly/3ZuzRgU

18

I’m looking forward to joining fellow healthcare and AI leaders at AI Summit 2026, hosted by the AI for PI Community, on October 6–7 in Atlanta. 👉 https://lnkd.in/egHcpMSZ As AI adoption accelerates in Payment Integrity, the differentiator is no longer model novelty — it’s reliability in production. ML, GenAI, and agentic AI only create value when their behavior is consistent, governed, and trusted in real-world #healthtech operations. From my perspective as a VP of AI, the goal is to turn advanced models into operational systems: * AI that runs before payment, not just after * GenAI that augments investigators and clinical reviewers, not replaces them * Models that are explainable, auditable, and regulator-ready * Architectures that scale across payers without sacrificing governance That’s where #ReliableAI becomes foundational. Reliability is what allows AI — and especially agentic systems — to act upstream with confidence. Codoxo AI rests on four pillars: 1. Compliant — secure, certified, and regulator-ready across data, models, and infrastructure 2. Explainable — interpretable outputs that enable confident human decision-making and next-best actions 3. Fair — embedded bias controls with continuous monitoring over time 4. Auditable — full traceability across model versions, tests, and outcome logs As agentic AI begins to coordinate workflows and decisions autonomously, these pillars become non-negotiable. Without reliability, autonomy amplifies risk; with it, agentic AI becomes a force multiplier for prevention-first operations. This Summit brings together leaders building real AI systems in production — not demos — and shaping what the next era of Payment Integrity will look like. Looking forward to the conversations ahead. #HealthcareAI #HealthTech #DigitalHealth #AIinHealthcare #AI4Good #Data4Good #GenerativeAI #AgenticAI #MachineLearning #ResponsibleAI #ReliableAI #PaymentIntegrity #AILeadership

40

I’ve been seeing increased focus on Model Validation in recent FRAML RFPs. What stands out is that many of the challenges are not technical. They stem from unclear ownership between Compliance, Technology, and Model Risk functions. Model Validation tends to work best when responsibilities are defined early and validation is treated as part of the operating model, not a downstream checkpoint.

7

Recently, someone asked who my AI hero is. After a little thought, it came to me: Rodney Brooks. What resonates for me is Brooks' insight that you don't need to mimic human cognition to create intelligent systems. Build systems that interact with their environment and let intelligence emerge from that interaction. Ant colonies are a natural example of this principle. They're smart enough to find the optimal food source despite individual ants having a tiny fraction of our neurons. In the 1980s, Brooks applied these principles to robotics using what he called the subsumption architecture. Each layer handled a simple task: avoid obstacles, follow walls, explore spaces. Stack them together, and complex behavior emerged. Intelligence from the bottom up. Three principles from Brooks' work continue to shape how I approach using today's AI at work: 1. Embodied. Brooks' robots were grounded in the physical world through sensors and actuators. When we apply this thinking to business and knowledge work, embodiment means something different. Data, information, and knowledge from enterprise systems, the internet, and people becomes the environment. That grounding is how we make today's AI useful. 2. Engineered. The subsumption architecture wasn't accidental. Brooks deliberately designed simple layers that could be composed. Same principle now: build modular capabilities that can combine and adapt. Claude's Skills system follows this pattern: each skill is a self-contained capability that agents discover and load only when relevant to the task. The intelligence comes from composition, not complexity. 3. Emergent. Brooks showed that some forms of intelligent behavior arise from layered, interacting systems. In an enterprise setting, this means creating systems that learn from use. When a system helps someone complete a task, captures what worked and what didn't, and applies that learning to similar tasks later, capability grows through interaction. You're not programming outcomes. You're designing conditions for improvement. Brooks' insights about embodied, emergent intelligence are still very much relevant today. I wonder how many businesses creating value with AI are following his principles.

37

3 Comments

We talk a lot about creating space for teams to innovate, build confidence, and feel safe to experiment. I believe deeply in that. At the same time, there’s an uncomfortable truth we don’t say often enough. Over time, teams can become comfortable. Comfort is not laziness. It is success optimized for stability. The problem is that technology innovation does not thrive in comfort. It thrives in learning, friction, and the willingness to challenge what worked yesterday. When leaders push teams to get comfortable with being uncomfortable, that pressure is often labeled as “directive.” I don’t think that’s accurate. Directive leadership removes agency. Productive discomfort expands responsibility. The difference matters. What complicates this further is that leadership influence is not only direct. In large organizations, PMOs, ITPMs, risk, and governance functions shape behavior quietly through tools, intake models, gates, and metrics. Even strong leaders who encourage autonomy can watch innovation stall because the system rewards predictability over learning. This is not a failure of intent. It is a design issue. Governance has a necessary role. Without it, scale collapses. But when governance drifts into command and control, teams stop thinking and start complying. The challenge for leaders is not choosing sides. It is tuning the system. We have to protect clear guardrails while questioning rigid “how.” We have to anchor conversations on outcomes, risk reduction, and value, not process loyalty. We have to create space where teams can experiment inside boundaries, not fight the system to move forward. The balance is never static. It shifts with context, maturity, and risk. Real leadership lives in that tension, pushing when comfort sets in and shielding teams when constraints overreach.

4