In this lightning round session, you will hear from the following speakers:

Platforming Temporal Cloud at Enterprise Scale

Jack Burns | Datadog

Rolling out Temporal across an enterprise means solving the boring problems: namespace provisioning, API key management, identity integration, and audit logging. This talk covers building a self-service platform that handles the operational work so teams can focus on workflows. We’ll share practical decisions around namespace design, SCIM integration, and infrastructure-as-code patterns.

Lessons learned from building fast, open source infrastructure for web agents

Catherine Jue | kernel.sh

Kernel builds fast, open source infrastructure that lets AI agents access the internet. To hit our sub‑150ms browser boot times, we evolved our stack from containers to unikernels to microVMs. This talk walks through what we learned with each iteration, and what it actually takes to make serverless browsers not just fast, but reliable at scale with long-running “entity workflows” in Temporal.

Running Agents on Async Task Queues While Keeping Them Real-Time

Ryan Stevens | ZoomInfo

AI agents are easy to demo in a web server and hard to run reliably at scale. As our workloads grew more long-running, asynchronous, and operationally critical, we moved execution onto Temporal task queues and re-architected our platform around durable workflows and distributed workers.

In this talk, I’ll show how task queues became the backbone of our agent runtime: letting us absorb bursty workloads, recover cleanly from failures, and run complex multi-step agents outside the request lifecycle. I’ll also cover the key architectural tradeoff we had to solve—how to move execution off the web tier without sacrificing the real-time product experience—and how we paired Temporal with streaming updates to keep the system feeling live to end users.

You’ll come away with a practical view of when web servers stop being enough for agent execution, why task queues matter, and how Temporal can provide the durability and scale these systems need.