AI assistant for Sentry
PocketClaw turns raw Sentry errors into something you can act on. An error tracker tells you an exception is happening; PocketClaw tells you whether it matters. When a customer reports that checkout is failing, it finds the matching exception, counts how many users hit it, and ties the spike to the deploy that preceded it. New errors show up in your morning brief before a customer emails about them. It prepares the fix task and the reply, and nothing is filed until you approve it.
What it does with Sentry
Matches an error to a customer report
PocketClaw searches Sentry for the exception that matches a complaint and its timing, so you confirm a real error instead of trusting the customer's account of it.
Counts who is affected
One report becomes a real number: the affected users and the error rate. That number is what decides whether it interrupts your day or goes on the list.
Connects the spike to a change
PocketClaw lines up when the errors started with what shipped, and names the deploy or pull request that most likely caused it.
Catches new errors in your brief
A new or spiking error is surfaced in your morning brief, so you hear about it from PocketClaw before you hear about it from a customer.
Seen end to end
Sentry is one of the tools PocketClaw reads for from customer issue to root cause. That page walks the whole job end to end, across every tool it touches.
Common questions
- Does it resolve errors for me?
- No. PocketClaw investigates: it matches the error, counts the impact, and names the likely cause. It prepares the fix task and the reply, and you approve both. Shipping the fix is yours.
- Does it work with Datadog or New Relic?
- Yes. PocketClaw also connects to Datadog, New Relic, and PagerDuty, and reads whichever monitoring tools you have connected.
- How does it know which change caused an error?
- It reads the timing of the error against your recent deploys and commits from your connected code host, and names the change that lines up as the likely cause.