How Walmart Is Reengineering AI Delivery Speed - with David Glick of Walmart (18 min)
- Release date: 2026-03-17
- Listen on Spotify: Open episode
- Episode description:
Enterprise AI is outpacing the operating models built to support it, forcing leaders to reconcile rapid iteration with safety, governance, and real‑world scale. In this episode, David Glick, SVP of Enterprise Business Services at Walmart, examines how stopwatch‑speed prototyping, nano‑agent architectures, and evolving security processes are reshaping enterprise delivery. The discussion highlights shifts from monoliths to federated agents, faster iteration cycles, and the emerging need to build the machine that builds the machine. Executives shaping real AI outcomes are invited to contribute their lessons to a curated peer audience. Learn more at go.emerj.com/expert to be considered for a future 'AI in Business' episode. Align your brand with the executives defining the enterprise AI agenda—partner at go.emerj.com/partner
Summary
- ⏱️ Stopwatch Over Calendar: Enterprises stall on quarterly timelines; AI demands real-time iteration via rapid prototyping and OODA loops to cut rework and align with user needs instantly.
- 🔍 Code-Driven Governance: Automate compliance by having AI agents read code for security answers, making deep processes like SSP real-time instead of months-long paperwork.
- 🏃♂️ Super Agile Evolution: From Waterfall to Agile to daily ‘super agile’ check-ins, faster cycles reduce planning risks and enable higher throughput, de-emphasizing rigid prioritization.
- 🤖 Nano & Super Agents: Task-specific nano agents handle narrow problems efficiently; super agents route semantically to them, scaling to thousands without monolith chaos.
- 🏭 Agent Factories: Build machines that build agents to keep pace with model advances; embrace team duplication for speed over perfection to ensure progress.
Insights
How can enterprises switch from quarterly calendars to stopwatch metrics for AI projects without compromising safety?
Time: 2:10 – 4:20
Category: AI in Workforce DisruptionAnswer: David Glick explains that AI enables rapid prototyping in minutes or hours, using real-time user collaboration to tighten OODA loops and reduce rework, while maintaining security through automated agents that scan code for compliance instead of manual processes. (Start at 2:10)
What if prototyping AI tools side-by-side with users could eliminate the gap between what users want and what engineers build?
Time: 2:36 – 3:26
Category: AI-Driven Innovation Economy, AI in Everyday LifeAnswer: The ‘swing set meme’ illustrates common misalignments in traditional development; real-time prototyping allows instant adjustments like moving buttons or adding workflows, accelerating user experience validation from months to the same day. (Start at 2:36)
Can AI agents make governance real-time by directly reading code to answer compliance questions?
Time: 3:47 – 4:20
Category: AI Governance & LawsAnswer: Instead of humans creating diagrams or filling forms, agents analyze code for databases, threats, and compliance, evolving security processes from months-long reviews to instant checks, keeping policies aligned with actual implementations. (Start at 3:47)
Is ‘super agile’ – with daily check-ins and hourly iterations – the evolution enterprises need beyond Waterfall and two-week sprints?
Time: 5:05 – 6:55
Category: AI in Workforce DisruptionAnswer: Faster iteration reduces the impact of planning errors; even as dates remain important for stakeholders, higher throughput (e.g., 50 projects/year vs. one) makes precise prioritization less critical, boosting overall delivery. (Start at 5:05)
Why are nano agents – single-developer, task-specific AIs – superior to monolithic systems or even microservices in massive enterprises?
Time: 7:37 – 9:03
Category: AI-Driven Innovation Economy, AI in Everyday LifeAnswer: Like specialized utensils over a Swiss Army knife, nano agents solve narrow problems quickly without coordination overhead; Walmart scales to thousands, moving from hundreds of engineers on monoliths to solo iterations. (Start at 7:37)
How do super agents function as semantic routers to connect users to the right nano agents without needing secret codes?
Time: 9:03 – 10:49
Category: AI-Driven Innovation EconomyAnswer: Super agents act as receptionists, using natural language to direct requests (e.g., revenue reports or lost cards) to domain-specific nano or chunky agents, simplifying intranet navigation at Walmart scale. (Start at 9:03)
What if the future of enterprise AI is ‘agent factories’ that automatically build customized agent-building machines?
Time: 14:30 – 15:44
Category: AI-Driven Innovation EconomyAnswer: Rapid model advances (e.g., rebuilding agents in days) demand meta-automation; Walmart experiments with multiple team-specific factories, prioritizing working prototypes over perfect standardization to avoid zero-output scenarios. (Start at 14:30)