OpenSRE provides an open-source AI agent that automates production maintenance, from incident triage to root cause analysis and fix delivery. The platform connects observability, code, and chat tools to run pre-built maintenance skills directly on a company's own infrastructure. It is designed to reduce on-call workload by resolving issues before engineers are paged.
Funding
Funding not disclosed
Founders
Product
Problem
Production engineering teams lose significant time manually investigating alerts, correlating telemetry across disconnected systems, and patching issues under on-call pressure. This reactive workflow delays fixes, increases operational burden, and diverts engineers from building long-term solutions.
Solution
OpenSRE offers a completely open-source AI SRE agent that automates production maintenance end-to-end. The platform schedules maintenance work and runs it directly in production, covering the full lifecycle from first signal to shipped fix and post-incident report. It handles CI fixes, security patches, and incident triage in one unified engine, tracing root causes, opening pull requests, and writing summaries automatically. The agent is opinionated rather than open-ended, using pre-built skills that connect to a company's observability, code, and chat systems. All operations are self-hosted under Apache 2.0, ensuring no production context leaves the customer's infrastructure and no vendor lock-in.
Target Audience
Primary customers are production engineering and SRE teams at software companies that need to reduce on-call toil and automate incident response, particularly those in data-intensive or bioinformatics environments with complex workloads.
Features
- Pre-built skills for production maintenance, including CI fixes, security patches, and incident triage, all running on the user's own infrastructure
- Automated root cause tracing, fix generation, and PR creation that ships resolved issues before engineers are paged
- Self-hosted deployment under Apache 2.0 with full transparency into every step the agent takes
- Integration layer that connects observability, code, and chat tools in approximately two minutes for live operation
- Proactive flagging of high-impact changes, such as a recent PR that lifted conversion by 6%
- Flaky pipeline detection and repair, including reruns, cause identification, and PR submission