Can I use it during an incident, not just afterwards?

Yes, and that is where it earns the most — assembling a timeline is the job nobody has hands for while they are also fixing the thing. Just do not put it in the decision path: it reads and reports, and every action stays with whoever is on call.

Does it understand PromQL?

It writes queries against your data sources, so what it can answer depends on what is actually recorded. Where a metric was never collected, no amount of querying invents it — an agent that says "no data for that window" is telling you something about your instrumentation.

How does this differ from connecting Sentry?

Sentry knows an exception was thrown; Grafana knows what the system was doing at the time. Connecting both is the combination worth having — the error and the conditions around it are what turn a stack trace into an explanation.

Observability

Grafana + Burrak AI

Get the story out of the dashboards before the incident review.

Connects with OAuth — revoke any time

Grafana is where the data is and rarely where the answer is: the panel shows a spike, and working out what preceded it means opening six more. An agent can search the dashboards, query the metrics behind them and read the logs alongside, then write the sequence out in order — which is most of what an incident timeline is, and all of what a post-mortem needs before anyone can argue about causes.

In the connector catalogue: Search dashboards, query metrics, explore logs.

Things you could ask for

Written the way you would actually say them. An agent plans the steps itself.

“Latency spiked at 03:10. Pull the metrics and logs around it and write the timeline.”
“Which dashboards actually cover this service? I inherited it and do not know.”
“Compare this week’s error rate with the same week last month and say whether it is drift or a step change.”
“Read the logs around this deploy and tell me whether anything degraded that no alert caught.”
“Every morning, summarise overnight anomalies and say which are worth looking at.”

It reads — dashboards, metrics and logs. It does not silence alerts, edit dashboards or acknowledge incidents, which during an incident is exactly the right division: the agent assembles the picture, the on-call decides.

Connecting it

1Create a Burrak account — the free plan is enough to try this.
2Open Connectors and choose Grafana.
3Authorise Burrak in Grafana's own window. Your password never reaches us; we receive a token you can revoke whenever you like.
4Give an agent a task. It will use the connector when the work calls for it.

Start free How connectors work

Grafana and Burrak, in practice

Can I use it during an incident, not just afterwards?: Yes, and that is where it earns the most — assembling a timeline is the job nobody has hands for while they are also fixing the thing. Just do not put it in the decision path: it reads and reports, and every action stays with whoever is on call.
Does it understand PromQL?: It writes queries against your data sources, so what it can answer depends on what is actually recorded. Where a metric was never collected, no amount of querying invents it — an agent that says "no data for that window" is telling you something about your instrumentation.
How does this differ from connecting Sentry?: Sentry knows an exception was thrown; Grafana knows what the system was doing at the time. Connecting both is the combination worth having — the error and the conditions around it are what turn a stack trace into an explanation.