Cisco President & CPO on Building Security-by-Design in AI Products | Jeetu Patel | E277

What It Covers

Jeetu Patel, Cisco President and CPO, explains how AI's shift to autonomous agents requires rethinking security architecture, organizational design, and product development across infrastructure, trust, and data layers at enterprise scale.

Key Questions Answered

• •AI Infrastructure Constraints: Training models now requires multiple data centers operating as one computer, with packets traveling hundreds of kilometers requiring MACsec and IPsec encryption baked into silicon to secure distributed training runs across geographically separated facilities.
• •Model Security Validation: AI models are nondeterministic and unpredictable, requiring runtime enforcement guardrails. Cisco jailbroke DeepSeek with 100% attack success rate on top 50 HARM Bench benchmarks, demonstrating why developers need third-party security layers to prevent prompt injection attacks.
• •Platform Organization Over Business Units: Conway's Law applies—organizations ship their org charts. Cisco eliminated divisional GMs to create one functional platform organization, reducing marginal cost of technology ingestion and creating compounding effects through loosely coupled, tightly integrated products across the portfolio.
• •Open Ecosystem Integration Strategy: Integrate with competitors including Microsoft, Zoom, Palo Alto Networks, and CrowdStrike. Product leaders must ensure SaaS applications surface in ChatGPT interfaces or risk exclusion as platforms consolidate. Partner broadly rather than building walled gardens to protect customer investment.