ChatOps brings DevOps operations into team chat, making them visible and collaborative. AI-powered ChatOps elevates this with natural language interfaces. Architecture: Slack Events API -> FastAPI webhook handler -> LLM tool-calling agent -> tool execution (Kubernetes, PagerDuty, Datadog, GitHub) -> Slack response. Slack app setup: create app at api.slack.com, enable Events API with request URL, subscribe to message events in channels. Authentication: OAuth 2.0 for workspace installation. Slash command implementation: /deploy -> parses intent (service, environment, version) -> validates permissions -> triggers deployment -> posts progress updates. LLM agent for complex queries: "What services are currently degraded?" -> agent queries Datadog metrics API + PagerDuty incidents + Kubernetes pod status -> synthesizes response. On-call automation: "Who is on-call for the payments service?" -> queries PagerDuty API -> formats response. Incident management: "Declare P1 incident for checkout service" -> creates PagerDuty incident, creates incident channel, invites on-call team, posts runbook link. Security: validate Slack request signatures, implement RBAC (only SREs can trigger production deployments), audit log all actions, require explicit confirmation for destructive operations. Implementation stack: Slack Bolt for Python, LangChain for LLM orchestration, Modal or Railway for hosting.