SynapseOasisSynapseOasisPrivacy
DocumentationSupport
Privacy policies/AI Portal Chat for JSM
Jira Service Management
Documentation for this app →

AI Portal Chat for JSM — Privacy policy

What this app stores inside your Atlassian tenant, what it only reads in memory, who it shares data with, and how everything is deleted.

Last updated: 28 August 2026 · App: AI Portal Chat for JSM · Vendor: SynapseOasis

This policy explains what data AI Portal Chat for JSM collects, how it is processed and stored, and the choices available to you. AI Portal Chat for JSM is an AI assistant for the Jira Service Management customer portal. It answers customer questions, searches your knowledge base, and creates or updates requests from the conversation.

AI Portal Chat for JSM is built on Atlassian Forge. It runs inside Atlassian's infrastructure and stores its data in your own Atlassian tenant. It calls no external service other than the Atlassian APIs of your own site.

1. Data we collect

The data falls into two categories: configuration and results held in app storage, and Atlassian data that is read and processed in memory but not retained.

App storage (persisted inside your Atlassian tenant)

  • Service desk and request type configuration — which desks and request types the assistant may use, whether knowledge base search is enabled and for which Confluence spaces, auto-confirm behaviour, and the instructions written by your administrators.
  • Conversations — the messages exchanged in a chat session, its status, the service desks and Confluence spaces the session was allowed to search, timestamps, and the thumbs up or down a customer may leave.
  • Branding — the trigger layout, colours, avatar name and the avatar image you upload.
  • An API token generated by the app, used to authenticate calls to its web trigger for managing operational notices. It is shown once and can be regenerated at any time.
  • Usage statistics — daily counters such as conversations, requests created and deflections, for the admin dashboard.
  • Audit entries — a record of the automated actions the app performed, with timestamps.
  • Per-issue AI results — the classification or suggestion produced for a request, stored against the issue key.
  • Agent skill profiles — a per-agent skill summary derived from previously handled requests, stored against the Atlassian account ID.

Transient data (read and processed in memory, not retained)

  • portal requests, issue fields, comments and request type metadata read from Jira Service Management
  • Confluence and knowledge base search results, limited to the spaces configured for that desk
  • files a customer attaches during a conversation, which are passed to Jira as request attachments
  • user, customer and group lookups from Jira

Transient data is read through the standard Atlassian REST APIs, used to produce the result you asked for, and then discarded.

2. Personal data

  • Chat messages are stored as written. If a customer types personal data into the chat, that text is retained in the conversation record until the conversation is deleted or the app is uninstalled.
  • Atlassian account IDs are stored when the app records an agent skill profile, an audit entry, or an action it performed on a request. They are the identifiers Atlassian itself uses and are meaningless outside your tenant.
  • Attachments are not retained by the app: they are uploaded to the Jira request and then follow that issue.
  • Once a day the app sends Atlassian the account IDs it holds, through Atlassian's Personal Data Reporting API, and Atlassian answers with the accounts that have been closed. Nothing else tells an app that a person left. Both stores are read and both are erased: in the key-value store a closed account's skill profile and the records keyed on it are removed, and in the audit table its rows are deleted — the row itself, not just the name on it, because an audit row carries an open metadata record about what happened.
  • The app does not collect passwords, end-user credentials or payment data.

3. AI processing

The AI features of this app run on Forge LLM, the AI runtime Atlassian provides inside the Atlassian platform, using Atlassian-hosted models. SynapseOasis operates no AI infrastructure of its own, holds no API keys to third-party AI providers, and never sends your data to one. Prompts are processed by Atlassian's service and are not used by SynapseOasis to train any model.

The following data is included in prompts when an AI feature runs:

  • the customer's chat messages and the history of the current conversation
  • request type names, field names and descriptions of the service desks the assistant is allowed to use
  • excerpts of Confluence pages and knowledge base articles returned by the search the assistant runs on the customer's behalf
  • the global, service-desk and request-type instructions written by your administrators, and any active operational notice

Model responses produce the result shown in the app and, where the app records a decision, are stored as described in section 1. AI features run only when enabled by your administrators.

4. Storage and retention

All persisted data is stored in Forge app storage (key-value store) and Forge SQL, both hosted by Atlassian, provisioned for your installation and located in the Atlassian cloud region of your site. SynapseOasis operates no servers, no databases and no logs outside Atlassian, and has no standing access to your data.

Configuration is retained for the life of the installation. Job data and results are retained until you delete or reset them, until they are replaced by a newer run, or until the app is uninstalled.

5. Data sharing

  • SynapseOasis does not share your data with third parties.
  • There is no sale or transfer of data to third parties.
  • There are no analytics, tracking or telemetry calls to external services.
  • The only outbound calls the app makes are to api.atlassian.com — the Atlassian REST APIs of your own site.
  • All data stays within Atlassian.

6. Security

The app runs inside the Atlassian Forge sandbox and is subject to Atlassian's platform security controls. It requests only the scopes it needs:

  • read:servicedesk-request
  • write:servicedesk-request
  • manage:servicedesk-customer
  • read:jira-user
  • read:jira-work
  • write:jira-work
  • search:confluence
  • read:space:confluence
  • storage:app
  • report:personal-data

Where the app acts on behalf of a user, Atlassian's permission model applies, so it cannot show a user data they could not already see. Administrative functions are restricted to users holding the corresponding Atlassian administration permission. All stored input is validated and size-capped before it is written.

7. Data deletion and retention

Deletion is controlled by the Atlassian platform, not by SynapseOasis. This section describes what actually happens, because the difference matters for a data protection assessment.

  • While the app is installed, you can delete data through the app's own screens, and that deletion is immediate.
  • When you uninstall the app, it runs an uninstall handler that clears its storage. Independently of that, Atlassian detaches the installation's data: it becomes inaccessible to the app, to your users and to us, right away.
  • After uninstalling, Atlassian keeps the detached data for a limited period before destroying it, under Atlassian's own data retention policy. Atlassian's Forge documentation describes the storage as soft deleted on uninstall and retained for 28 days, and separately notes that a re-link request must reach Atlassian within 21 days for previous data to be restored to a new installation. That restore only happens if you ask us to raise the request, with your consent. We never initiate it.
  • Atlassian's backups follow Atlassian's own schedule and are outside any app's control.

The practical consequences: we cannot delete this data faster than Atlassian's process allows, and we cannot read it after an uninstall. If you need specific data gone on a specific date, delete it inside the app before you uninstall.

Atlassian documents this in Data lifecycle for Forge-hosted storage. The retention periods are Atlassian's to change, so treat Atlassian's documentation as the current source rather than this page.

8. GDPR

  • Data minimisation — only the configuration and results needed for the app's function are stored; everything else is processed in memory and discarded.
  • Purpose limitation — data is used solely to provide the app's functionality inside your tenant, and is never sold or transferred to third parties.
  • Data residency — all persisted data remains in your Atlassian tenant on Forge infrastructure, in the Atlassian cloud region of your site.
  • Right to erasure — data can be deleted through the app while it is installed. On uninstall it is detached immediately and then destroyed by Atlassian under Atlassian’s retention policy, as described in the section above.
  • Sub-processors — SynapseOasis uses no sub-processors for app data. Atlassian is the infrastructure provider and processes the data under your existing agreement with Atlassian.

9. Contact

10. Changes to this policy

We may update this policy from time to time. When we do, we revise the last-updated date at the top of this page. We encourage you to review it periodically to stay informed about how AI Portal Chat for JSM handles your data.