Groundcover Investigations lets engineers steer AI SRE agents
Groundcover, the eBPF-based observability startup, has introduced Investigations, giving agents and engineers one place to collaborate on incident response in The post Groundcover Investigations lets engineers steer AI SRE agents appeared first on The New Stack .
Groundcover, an eBPF-based observability startup, has developed Investigations, a tool that enables engineers and AI SRE agents to collaborate on incident response in real time. This shared workspace allows agents and engineers to work together to investigate production incidents, combining their strengths to create a shared record of evidence.
According to Shahar Azulay, CEO and co-founder of Groundcover, most AI SRE tools ask engineers to step back and let the agent take over, but Investigations is designed for engineers to step in and steer the investigation. The tool aims to eliminate fragmented incident response and create a single source of truth for every engineer and agent.
When an issue is detected, an investigation can be opened from Groundcover or Slack, and the investigator agent builds a graph to test hypotheses against production data. The investigator agent submits proposals for each hypothesis, which are then checked by a separate agent, the moderator, using LLM-as-a-judge to ensure the evidence supports the claim.
This separation of hypothesis and conclusion helps to ensure trustworthy conclusions. Groundcover's investigator and moderator agents lead investigations, but engineers can also connect their favorite coding agents, such as Claude Code or Codex, to the shared workspace. These external coding agents can query telemetry and hand off investigations to Groundcover's own agents.
The separation of work and judgment allows each agent to shine in what they do best, and every agent's findings are published in the same investigation tree, making it easy for engineers to see what has been checked, what is still open, and why each conclusion was reached.
Written by urgent.news from The New Stack's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.