Upstream: Vendor Outage Detector for Jira Service Management

Is it Stripe or us? Upstream joins your vendors’ status pages with a spike in your own requests and answers in seconds — then puts up the portal banner, replies to the linked requests and closes out when the vendor recovers.

  • For Jira Service Management Cloud
  • Free up to 10 users
  • Built on Atlassian Forge
  • Reads public status pages only — nothing of your desk is sent to them
On this page
  1. One verdict: vendor or us
  2. The whole incident loop in one click
  3. A detector that knows your normal
  4. A vendor ledger for procurement
  5. An agent in Rovo
  6. Try it in 20 seconds
  7. FAQ

Upstream answers the first question of every outage — is it our vendor or us? — in seconds. It reads your vendors’ status pages, watches your own customer requests for a spike, and joins both into one verdict. Then it runs the customer side for you: a banner on the portal, one reply to every linked request, and a closing comment when the vendor recovers.

One verdict: vendor or us

The Live tab shows every vendor your desk depends on, its status page right now, and a verdict in plain words: Twilio is degraded, your desk is quiet — or Stripe is down and your payment requests are spiking. A spike with no vendor behind it shows as yours.

The whole incident loop in one click

When a vendor is down, the Incidents tab opens a Jira incident and lists four steps: open the incident, put a banner on the portal for the affected request types, send one reply to every linked request, and close out when the vendor’s page says resolved. Do all four runs them; you approve the texts first.

A detector that knows your normal

The Detector learns how many requests each request type gets at the same hour of the week. When requests rise above that band, it marks a spike within minutes — often before the vendor’s status page says anything. Vendor keywords and bindings tell it which request types belong to which vendor.

A vendor ledger for procurement

The Vendor ledger adds up the year: each vendor’s incidents and hours degraded, the requests they caused, the agent time they cost and the service credits you can claim under their terms. Export for procurement before the renewal.

An agent in Rovo

The Upstream agent in Rovo answers “is it them?”, drafts customer comms and opens an incident when you ask. On sites with Rovo it works next to the app; a site admin can also switch on the MCP tools for an AI client.

Try it in 20 seconds

Replay the Stripe incident on the Get started tab: a recorded outage from the first customer request to close-out, with the verdict, the banner and the reply. Nothing reaches a customer.

Frequently asked questions

Which vendors does Upstream follow?

Stripe, AWS, Duo, Atlassian, Twilio, GitHub, Cloudflare, Google Cloud and Azure out of the box, plus up to ten status pages of your own (any Statuspage page, such as DigitalOcean or npm) that a site admin approves.

Does it send anything about my desk to the status pages?

No. Upstream only reads public status pages. No request text, customer name or email leaves Atlassian; see the privacy policy.

Does it post to customers on its own?

No. A lead approves the banner, the reply and the closing text. Do all four runs the steps a person chose; Upstream drafts the texts and you can edit them first.

How does it know a spike is not just a busy day?

The Detector learns each request type's normal level for the same hour of the week over four weeks. A spike is when the requests rise above that band; you set how sensitive it is.

Can I try it without an outage?

Yes. Replay the Stripe incident on the Get started tab shows a recorded outage from spike to close-out in 20 seconds. No customer sees it.

Get Upstream

Free up to 10 users. Install it from the Atlassian Marketplace.