Day 2 · Recorded 7 May 2026
How incident.io designs background agents people actually work with
The hard part is not just giving a background agent a channel. It is making the agent's work fit the rhythms of responders in Slack, engineers in IDEs, and other agents already paired with those humans. Lawrence Jones uses incident.io's AI SRE to show what real collaboration feels like when the agent is part of the team workflow.
Lawrence Jones, Founding Engineer, incident.io
What's in this session
Background agents do not create value just by producing output. In incident response, their work has to be visible, timely, trusted, and available wherever responders are actually working.
Lawrence Jones walks through AI SRE as a worked example: pinned Slack summaries, multiple summary formats, full-investigation canvases, threaded progress updates with citations, rules for proactive heads-up messages, IDE bridges through MCP tools, and Investigation Sync for moving incident context into local agent workspaces.
For product, platform, and engineering leaders, the lesson is to design the handoff surface as carefully as the agent itself. Clear boundaries, native channels, real-time bridges, and high-fidelity agent-to-agent handoffs turn background agents from isolated automation into collaborators.
Inside the recording
- 00:00 AI SRE as a worked example
Lawrence frames background agents as collaborators with humans and their paired local agents. - 04:00 Channels are the agent interface
Choosing Slack creates constraints around message length, visibility, and how responders consume work. - 09:00 Designing summaries people actually use
Incident responders need skim-friendly updates, deeper canvases, and formats shaped by research. - 12:00 Progress messages with citations
AISRE formats every check consistently so people can read, trust, and inspect the work. - 16:00 Be where engineers work
Slack is not enough, so incident context syncs into IDEs and local agent workspaces. - 20:00 Handoff to delegated coding agents
Central specialist agents need clean transfer paths into the code agents teams already use.
More sessions on agent infrastructure
- Spectre: Harvey's collaborative cloud agent platform — Joey Wang, Harvey
- Backgrounding the Toil: Inside Uber's Agent-Ready Developer Platform — Nikhil Ramakrishnan, Uber
- Enabling AI Tools Without Losing Control: Lessons from a Regulated Bank — Suhail Patel, Monzo