How Tanzu Ships Software with AI: 15+ Agents Across the SDLC | Purnima Padmanabhan

Purnima Padmanabhan, GM of the Tanzu Division at Broadcom, sits down with @thecote to talk about what's actually changed in how her team ships enterprise software now that AI is woven through every phase of the development lifecycle - and what shipped in Tanzu Platform 10.4 to help customers do the same. We cover: Why code generation is only 20-30% of the SDLC, and where the real acceleration comes from Finding (and fixing) chained low-severity vulnerabilities faster than attackers can Restructuring teams into pods, and the "challenger" role that keeps humans from rubber-stamping beautiful-but-wrong AI output Using Spring and frameworks to dial in the balance between AI creativity and determinism The 15-20 agents now running across requirements, prototyping, coding swarms, testing, docs, security, and packaging Tanzu Platform 10.4's Agent Foundations: a secure agent runtime that is deny-by-default for network, secrets, and resources The Alibaba "Rome" agent that decided to mine Bitcoin - and what that teaches us about agent boundaries Agent Buildpacks, Spring AI, MCP gateway brokering, and how `cf push` gets an agent running in minutes Read more about Tanzu Platform 10.4: https://blogs.vmware.com/tanzu/introd... Chapters: 00:00 Shipping with AI trust 02:15 Security patches accelerate 03:30 Feature cadence balance 05:07 AI across the SDLC 08:13 The adoption tipping point 09:38 Key lessons learned 16:05 Org changes and the skeptic role 18:34 Frameworks for consistency 20:11 Agents everywhere now 22:30 Prototyping and metrics 27:34 What Tanzu Platform does 33:09 Secure agent runtime 36:47 Agent Foundations toolkit 41:16 Release cadence going forward 43:46 Wrap up Learn more about Tanzu Platform: https://tanzu.vmware.com