Incidents fixed while you sleep.
Humans firmly in charge.
Agent 99 is the AI-native operating system for SRE & DevOps: one place where signals correlate into incidents, agents propose fixes, and every action is approved, audited, and reversible.
Signs in with your Chibitek account. AI is advisory by design — no autonomous action without a human decision or an earned, bounded policy envelope.The Agent 99 lifecycle
- 1Monitor
Signals and metrics from every source land in one place — HMAC-signed ingestion, SIEM bridge, HTTP checks.
- 2Detect
Deterministic correlation folds repeat signals into one incident per (service, fingerprint) — noise becomes signal.
- 3Investigate
Timeline, correlated signals, golden signals, and MTTA/MTTM/MTTR from your real history.
- 4Respond
Acknowledge → investigate → mitigate → resolve, with reversal for mistakes and blameless postmortems.
- 5Automate
An AI agent fleet proposes; you approve. Trust scores, an autonomy ladder, and an append-only ledger.
Built on trust, not vibes
Incident Response
Correlation engine, full lifecycle with reversal, on-call rotations, postmortems drafted from the real timeline.
Observability
Monitors evaluated every 5 minutes, golden signals (P50/P95/P99), SLOs with availability and burn-rate tracking.
AI Agents, governed
Shadow → approval-gated → autonomous. Risk-scored approval cards, bounded reversible actions, trust that is earned.
Nova Shell
Natural language to a governed command catalog. Every dry-run audited; write-risk commands gate on approval.
API, webhooks & CLI
Scoped bcrypt-hashed tokens, HMAC-signed outbound webhooks, and the agent99 CLI for your terminal.
Enterprise identity
Chibitek Stiki SSO (RS256), per-client database isolation, RLS, soft deletes, and an immutable audit trail.
Every action accountable
Append-only timelines. Bcrypt-hashed tokens. HMAC-signed ingestion and delivery. Row-level security in a per-client database. An agent ledger where trust is earned two points at a time — and lost four at a time.
Read the trust page