Background Agents Virtual Summit 2026
2026-05-06..2026-05-07
-- online --
who it's for
platform engineers, site reliability engineers, ai/agent engineers, security engineers, engineering managers, developer experience engineers, data platform engineers
topics
cloud dev boxes and devpods; kernel-enforced sandboxing (Nono); agent runtime security and attestations; AI code reviewers for CI gating; ephemeral development environments and warm snapshots; agent orchestration and fleets (Ona, Specter); Sourcegraph code intelligence integration; deterministic blueprints for PR automation; Merkle audit logs and syscall recording; registry for versioned agent context; Slack-thread collaborative cloud agents
| title | author | length |
|---|---|---|
| What Comes After AI Coding Assistants | Will McMullen, Philipp Pietsch | 16m |
Background agents move code generation off laptops into the cloud and claim to free teams from manual initiation by combining ephemeral development… | ||
| Minions: One-Shot Coding Agents on a 30M-Line Codebase | Alistair Gray | 38m |
Stripe Minions convert Slack, Jira, or web requests into branch + PRs on a 30M-line Ruby (non‑Rails) codebase, producing thousands of agent‑authored PRs merged… | ||
| AI Engineering Stack: Assisted to Delegated | Rajesh Bhatia | 38m |
Cloudflare reports 93% R&D adoption after wiring AI Gateway for model routing, Access for identity, Backstage as a service catalog, and Agents.md for agent… | ||
| Agent Runtime Security: Enforce, Attest, Decide | Stephen Parkinson | 28m |
Nono is a CLI kernel-enforced sandbox that wraps any agent and guarantees the kernel-enforced cage cannot be bypassed, using composable policy groups in… | ||
| Internal Background Agent Systems: Architecture and Rollout | Cole Murray | 31m |
Open Inspect runs a Cloudflare-based control plane and data plane split with sandboxes on Modal and Daytona, aiming for reproducible dev environments via warm… | ||
| Background Agents for Genomics and Cloud Ops | Xiucheng Quek | 23m |
Genomics platform runs at exabyte scale on AWS and uses background agents to triage runs, write incident reports, and merge safe code changes into main after… | ||
| Why Prompt-Level Guardrails Fail for AI Agents | Leo Di Donato, Lorenzo Fontana | 30m |
Claim: prompt-level guardrails fail once autonomous agents can run tools, read files, adapt and self-modify, so sandboxing or microVMs on a laptop is… | ||
| Context Is the New Code for AI Agents | Patrick Debois | 15m |
Argues that context needs a development lifecycle with concrete practices: linting and testing prompts and skills using unit-like task files and using LLM as… | ||
| Day 1 Recap: What Practitioners Learned | Will McMullen, Philipp Pietsch | 16m |
Day 1 surfaces practical fixes: codified engineering standards that restore trust in AI review, dev boxes in VPCs for isolated agent environments… | ||
| Day 2: From Foundations to Production | Will McMullen, Philipp Pietsch | 12m |
Day two argues that concrete primitives win: reproducible dev environments like DevPod / cloud dev boxes make agents actually reproducible, while OpenInspect… | ||
| Collaborative Cloud Agents in Slack Threads | Joey Wang | 28m |
Harvey moves coding agents off laptops into a cloud platform called Specter that stores session state as JSONL, wires Codex/cloud code and MCPs into Slack… | ||
| AI SRE and Agent Handoffs: incident.io's Collaboration Model | Lawrence Jones | 24m |
Incident.io’s AI SRE runs live investigations in Slack, pins a short summary and updates a threaded progress feed, while a separate human-facing summarizer… | ||
| Software Factory in Public: 375 PRs in 10 Days | Zacharias Malguitou | 28m |
A public factory builds Memo with no human-written code and merges 375 PRs totaling ~68,000 LOC and over 1,000 tests in 10 days; metrics show 38 minute median… | ||
| Minion: Agent-Ready Dev Platform Behind 11% of Merged PRs | Nikhil Ramakrishnan | 15m |
Uber’s Minion platform generates 11% of merged PRs by plugging specialist agents into CI and remote DevPod environments to automate toil like tests, diffs, and… | ||
| Dark Factories: Verification-First Agentic Engineering | Shardul Vaidya | 20m |
Claims that noninteractive runs show a 53% accuracy decline and that 91% of leaders report velocity gains but only 25% have data drive a verification-first… | ||
| AI Tools in Regulated Banking: Monzo's Control Model | Suhail Patel | 45m |
Monzo runs an opinionated monorepo with 3,000+ microservices and uses static analysis, automated code mods and service generators to enforce auth and… | ||
| Wrap: What To Build Next | Will McMullen, Philipp Pietsch | 19m |
14 sessions showcase concrete choices: build isolated dev boxes in your cloud, enforce controls at kernel-level enforcement, and use data walls like VDO and… | ||
| nothing matches | ||