Now in early access

Your AI SRE,

Install a lightweight agent. Ask questions in plain English. OpsPilot reasons through the problem, runs diagnostics, and explains what's wrong — no 2am phone calls.

web01 — bash
# Install OpsPilot agent (Linux)
$ curl -sSL "https://cp.911devops.com/install/linux?api_key=YOUR_KEY" | sudo bash
✓ Verifying SHA256...
✓ Installing systemd service...
✓ Agent online — heartbeat active
1,695 tests passingLinux · macOS · Windows3-layer safety enforcementOllama · Claude · AWS BedrockJWT auth + TLSFull audit log
The problem

Sound familiar?

These aren't edge cases. They happen every week.

03:14 — ALERT

The 3am disk alert

You SSH in. Run df. Run du. Grep through logs. Forty-five minutes later you find a stale nginx log nobody rotated. You go back to sleep. It happens again next week.

OpsPilot solves this.

14:32 — ALERT🔴

The new on-call engineer

Something's broken on prod. They know it's broken. They don't know which logs to check, which commands are safe to run, or which service is downstream of what. They page the senior dev.

OpsPilot solves this.

09:00 — ALERT📋

The compliance audit

Auditor asks: show us every command run on prod last month. You have rsync timestamps and tribal knowledge. You have nothing.

OpsPilot solves this.

How it works

Three steps to AI-powered ops

From blank server to intelligent diagnostics in under two minutes.

01

Install the agent

One curl command. The agent registers itself, starts sending heartbeats, and is ready in under 60 seconds.

bash
$ curl -sSL "https://cp.911devops.com/install/linux?api_key=YOUR_KEY" | sudo bash

✓ Downloading agent v0.9.4...
✓ Verifying SHA256...
✓ Installing systemd service...
✓ Registered: web01.acme.com → control plane
  Agent online. Heartbeat active.
02

Ask in plain English

No commands to memorize. Describe the problem exactly as you'd say it to a senior SRE.

bash
opspilot > why is web01 running out of disk?

Thinking... analyzing web01
03

AI reasons and explains

OpsPilot runs diagnostic steps, observes each result, and returns a plain-English root cause with a recommended fix.

bash
→ Checking disk usage...          df -h     ✓
→ Finding large directories...    du -sh /*  ✓
→ Checking recent writes...       stat ...   ✓

⚡ Root cause identified:
   /var/log/nginx/access.log  38 GB
   Last modified: 8 days ago. No active writes.

Fix: truncate -s 0 /var/log/nginx/access.log
     (or configure logrotate — see /etc/logrotate.d/)
Built for production

Every feature ships working

Not a demo. 1,695 tests passing. Every capability is live today.

Plain-English diagnostics

Ask exactly what you'd ask a senior SRE. OpsPilot translates it into the right safe command sequence and explains findings in full sentences.

Safe by design

Hard-coded DENY list + YAML policy + model self-assessed risk. rm, kill, reboot, dd never run — regardless of what the AI suggests. Three independent layers.

Any AI backend

Local Ollama (air-gapped, nothing leaves your network), Claude API, AWS Bedrock. Switch in config.yaml — no code changes, no restart.

Full audit trail

Every command logged: actor, capability, risk level, duration, exit code. SOC 2 / HIPAA ready from day one. Query the log via API.

Cross-platform agents

Linux (systemd), macOS (launchd), Windows (NSSM). Same control plane. Same dashboard. Same API. One install command per OS.

Auto-updating agents

On startup and every hour, agents check for updates, download, verify SHA256, and restart in place. Zero manual updates across your fleet.

Pricing

Pay for what you use

Start free with up to 5 servers. Upgrade when you need cloud AI or more scale.

Free forever

Basic

Free
IncludedUp to 5 servers
Max5 servers max
OverageUpgrade to add more
AI backendaitower (local AI)
SupportCommunity
MFA / TOTP
Audit log
Avg. response~2–8s / query
Get started free →
Most popular

Pro

$79/mo
Included10 servers included
MaxUp to 50 servers
Overage$10 / server over limit
AI backendClaude Haiku (Bedrock)
SupportEmail — 48h SLA
MFA / TOTP
Audit log
Avg. response~1–3s / query
Get started →

Elite

$149/mo
Included25 servers included
MaxUnlimited servers
Overage$8 / server over limit
AI backendClaude Sonnet (Bedrock)
SupportPriority
MFA / TOTP
Audit log
Avg. response~2–5s / query
Get started →

Basic is free while we're in early access. Avg. response times are per-query estimates and vary with AI backend and server load.

Early access

Request early access

We're onboarding a small group of early customers. Fill this out and we'll send you an install token + a 30-minute onboarding call.

FAQ

Common questions

Only for cloud AI. With Ollama: fully air-gapped. The agent, the control plane, and the AI all run inside your network. Nothing leaves your infrastructure.

An explicit allowlist of ~30 read-only diagnostics: df, free, ps, ss, uptime, ip addr, lscpu, systemctl status, journalctl, netstat, and similar. rm, kill, reboot, dd are permanently blocked — no override possible.

Linux (Ubuntu 22/24, Debian 12, RHEL 8/9, Rocky 9, Amazon Linux 2023), macOS (Monterey+), Windows (Server 2019+, Win 10+). Same agent binary, same control plane.

Bootstrap API key on first install → control plane issues a per-agent signed JWT (HS256, 30-day expiry). Tokens auto-rotate. TLS enforced on all connections. No plaintext secrets on disk.

Yes. TOTP MFA (Google Authenticator, Authy, 1Password) with backup codes. Available on all accounts — managers, admins, and standard users.

Yes. On startup and every hour, the agent checks the control plane for a newer version, downloads it, verifies SHA256, and restarts itself. Zero manual updates across your fleet.

opspilot

Ready to stop guessing
what's wrong?

Install in 60 seconds. Get your first AI diagnosis in under 5 minutes. No credit card required to start.

No credit card·Cancel anytime·Installs in 60s