OneGrep

What is OneGrep?

OneGrep is a Y Combinator W24 startup that builds an AI DevOps automation agent for incident response. During incidents, engineers repeatedly perform the same investigation steps: check CloudWatch logs, query the database, look at recent deployments, check Kubernetes pod status. OneGrep automates these steps — executing runbook tasks across your entire tool stack automatically when incidents occur.

Key Features

Runbook Automation — Convert your existing runbooks into executable workflows. When an alert fires, OneGrep runs the investigation steps automatically: querying logs, checking metrics, executing AWS CLI commands, searching Slack for related discussions.

Universal Tool Integration — OneGrep connects to the tools engineers actually use during incidents: PagerDuty, OpsGenie, AWS, GCP, Kubernetes, Datadog, Grafana, Slack, GitHub, Jira, and more. It acts as an AI agent that can use all these tools simultaneously.

Natural Language Commands — During active incidents, engineers can instruct OneGrep in natural language: "check what changed in the last 30 minutes," "show me all errors from the payment service," "restart the affected pods." OneGrep translates these into the appropriate tool calls.

Institutional Knowledge Capture — OneGrep learns from how engineers resolve incidents and builds a knowledge base of resolution patterns, making future similar incidents faster to resolve.

Audit Trail — Every automated action is logged with full context — what OneGrep checked, what it found, and what actions it took — providing an automatic incident timeline for postmortems.

How It Differs from Shoreline

Shoreline focuses on pre-built automation runbooks that execute on triggers. OneGrep is more agentic — it can respond to natural language mid-incident and adaptively investigates based on what it finds, rather than following fixed scripts.

Bottom Line

OneGrep is a strong fit for engineering teams with well-documented runbooks who want to automate the execution of those runbooks without rewriting them as code. The natural language interface during live incidents is a meaningful UX improvement over traditional runbook automation.