Agents think in milliseconds, legacy infrastructure doesn't. LinkedIn, Walmart and Zendesk shared how they closed the gap at VB Transform 2026

At VB Transform 2026, infrastructure leaders from LinkedIn, Walmart, and Zendesk said the biggest obstacles to scaling AI agents came from legacy infrastructure, not model quality. They described systems built for human-paced workflows breaking down when agents needed millisecond-level orchestration, evaluation, and deployment.

The discussion signals a shift in enterprise AI work from model experimentation to infrastructure redesign. As companies move agents into production, competitive advantage may depend less on choosing the best model and more on building fast, reliable platforms that can support autonomous workloads at scale.