Triage incidents
Ranked causes for the symptom you describe.
Product
Your calm second opinion at 3am.
Vigil turns a symptom into a short, structured investigation: what is most likely wrong, how to confirm it safely, and what to do next. Every answer ends with follow-up questions, and your chats are saved in your browser to pick up later.
How it works
A method, not another guessing machine.
Vigil guides the investigation. Your engineers stay in control of every command and every decision.
01
Describe
Paste the symptom, error or alert, with names and secrets removed.
02
Investigate
Get ranked hypotheses and the read-only checks that separate them.
03
Decide
Run the checks yourself, share what you see, and choose the fix with your team.
Why not a general chatbot?
Answers shaped for on-call, not for essays.
General AI chat is great at many things. Under pressure, SRE teams need structure, safety and evidence.
General AI chat
Vigil
Long, generic essays
On-call-length answers: summary, likely causes, checks, next step
Guesses presented as facts
Hypotheses labelled and tied to the evidence that confirms them
May suggest risky fixes first
Read-only commands first; state-changing steps clearly flagged
Commands buried in prose
Copy-ready commands with placeholders for your resources
Use cases
Built for the moments that page you.
Pick a scenario to see how Vigil answers. Every reply is short, structured and ready to act on.
VigilExample answer
payments-api pods are in CrashLoopBackOff after the 14:02 deploy.
The container exits right after start; Kubernetes is backing off between restarts.
Likely causes
- Startup error from new config or a missing env var
- OOMKilled: memory limit too low (exit code 137)
- Liveness probe kills the app before it is ready
bashCopy
kubectl logs deploy/payments-api -n <namespace> --previous
Ask next
How do I compare the two releases?How do I roll back safely?
Capabilities
One copilot, the whole incident.
From the first alert to the follow-up tickets, Vigil helps at every step, with guardrails built in.
Read your logs
Attach a file; Vigil finds the lines that matter.
Explain alerts
What fired, why it usually fires, what to check.
Write queries
PromQL, Logs Insights and kubectl, ready to paste.
Suggest safe checks
Read-only first; risky steps clearly flagged.
Ask the next question
Every answer ends with one-tap follow-ups.
Draft postmortems
Timeline, factors and actions from rough notes.
Share with the team
Revocable links, Markdown and email.
No access to your infrastructure
Nothing to install, no credentials, no agents in your cluster. You share symptoms; it never touches production.
Read-only first
Every check starts with safe get, describe, logs and metrics commands. Anything that changes state is flagged.
Hypotheses, not guesses
Causes are ranked by how well they fit your evidence, each with the signal that confirms or rules it out.
Early access
Give your on-call team a calmer 3am.
Join the waitlist for team workspaces, integrations and private deployments.