|
May 5, 2026
[LIVE] Tomorrow: Stripe, Cloudflare, & more on running agents at scale
This was the last-call issue for the Background Agents Summit, running May 6 and 7. The framing was simple: agentic engineering had moved past the question of whether agents can write code and into the harder questions about trust, runtime design, review, and operational overhead.
The issue emphasized that even if you could not attend live, registration still mattered because all recordings would be sent afterward.
What This Issue Covered
|
|
Why this summit mattered now: teams had moved from asking whether agents can write code to figuring out where to run them, what to trust them with, and how to review their work.
Who was speaking: Stripe, Cloudflare, Harvey, Uber, AWS, Monzo, Open Inspect, Genentech, incident.io, software-factory.dev, and more.
How to attend: live on May 6 and 7, or via recordings after registration.
|
What Happens After Agents Can Write Code?
That was the central question in this issue. The point was not that code generation had been solved for everyone, but that leading teams were already dealing with second-order problems: runtime architecture, review flow, trust boundaries, context transfer, and how to make agent output increase leverage instead of creating more work.
Examples called out directly in the issue included Stripe's 1,300+ agent PRs a week and Ramp's Inspect handling more than half of merged PRs. The summit was positioned as a way for teams to learn how that kind of system actually works in practice.
Live
Two-day schedule highlights
Day 1: Stripe on operating agents on a 30-million-line codebase, Cloudflare on delegated AI engineering, Open Inspect on company-internal background agents, and Genentech on running agents across science and cloud infrastructure.
Day 2: Harvey on Spectre, incident.io on collaborating with background agents, software-factory.dev on building in public, Uber on an agent-ready developer platform, AWS on "Dark Factories," and Monzo on enabling AI tools in a regulated bank.
|
Agenda Themes
The published agenda centered on the topics teams actually struggle with once agents become production systems:
Context: how agents get the information they need to act effectively.
Review: how humans inspect, route, and trust higher volumes of machine-generated work.
Security and identity: how to constrain agents without making them useless.
Runtime design: how to move beyond local copilots and into repeatable background-agent systems.
Register
If your team was already experimenting with agents, this issue positioned the summit as the fastest way to sharpen instincts, compare architectures, and avoid mistakes other teams had already made.
Join the Background Agents Summit.
See you there,
Lou
Field CTO, Ona
|