How to use Velocity to investigate production issues, on demand or automatically.
Make sure the Velocity Slack app is installed and added to the Slack channel where you’re trying to use it. If Velocity isn’t in the channel, it won’t be able to respond or investigate.
This guide walks you through the three main ways to trigger Velocity investigations:
Reply to an existing message, alert, or thread.
Start a new message with context.
Automatically trigger investigations via alert and incident rules.
💡 Tips for Best Results
Be clear and specific when asking Velocity to investigate.
Include:
Service names
Timeframes
Relevant context or symptoms
Use Slack threads whenever possible for continuity and context.
1. 🧵 Investigate an Existing Slack Thread / Message
You can ask Velocity to investigate directly within an existing Slack thread, whether it’s an alert from a monitoring tool or an active conversation.
When to use:
You’re replying to an alert (e.g., from Datadog, PagerDuty, OpsGenie, Incident.io, Grafana Alerts, etc...).
You’re joining a conversation about a production issue.
How to trigger:
Reply in the thread with the relevant instructions to Velocity, for example:
@velocity investigate
@velocity ????
@velocity what’s going on here?
Velocity will reply in-thread with a detailed investigation summary.
Tip: Thread replies ensure Velocity has the full alert and discussion context.
2. 🆕 Start a New Investigation
You can trigger a new investigation by messaging @velocity in any Slack channel with the relevant context.
When to use:
You noticed something off (e.g., latency, error spikes).
You received an alert outside the normal alerting flow.
You want to proactively analyze service behavior.
How to trigger:
Send a message like:
@velocity I have an issue with service-x that started around 2 PM UTC. Please investigate.
@velocity analyze the last hour of dashboard-x and let me know if you see anything unusual.
@velocity I got this alert about 5xx errors from API-Gateway. Can you check?
@velocity CPU usage on service-y is spiking since 10 AM. What's going on?
The more specific your message (service name, timeframe, context), the better Velocity can help.
