How to replicate Atlassian Support “AI in form before ticket creation” pattern in JSM?

[Mariano Tallarico _Appricity_]

Body:
Hi everyone,

I’m a Jira / Jira Service Management administrator and consultant, and I’m trying to understand how to replicate a specific pattern I see in Atlassian’s own support experience.

On the URL https://support.atlassian.com/contact/#/, when you select the type of problem (for example “Technical issues and bugs”) and the app (for example “Jira Service Management (Cloud Premium)”), an AI text box appears inside the same form. There you can describe your issue and get an AI-generated answer before actually creating a ticket. Only if you still need help, you proceed with submitting the request.

In other words, the flow is:

1. User chooses topic and product.
2. An AI assistant appears inside the form.
3. User describes the problem and gets an answer (using KB / docs / etc.).
4. Only if necessary, the user escalates and creates a JSM request.

For my customers, I would love to implement a similar interaction model, with these goals:

• Have an AI assistant embedded directly in the “contact/help” experience (ideally inside a form).
• Let the user get answers without automatically creating a ticket.
• Create a JSM request only when the user explicitly requests it or when AI cannot solve the issue.
• Reduce the operational load on the service desk by resolving as much as possible before ticket creation.

My questions for the community are:

1. Is the AI form experience on https://support.atlassian.com/contact/#/ built on top of any standard Jira Service Management / Rovo functionality that is (or will be) available to Cloud customers and/or partners, or is it a fully custom implementation?
2. Today, is there any supported way to embed a similar AI chat/assistant inside a JSM customer portal form (request type), before ticket creation? For example:
   • Any Rovo integration specifically designed for JSM customer portals.
   • Any Forge module or extensibility point to add an AI “chat box” into the customer portal UI.
   • Any recommended architecture to implement a “pre-ticket AI layer” that still integrates cleanly with JSM.
3. If this is currently only possible with custom front-end + APIs, I’d really appreciate:
   • Examples or references of similar architectures that other customers/partners have implemented.
   • Any public feature request (jira.atlassian.com) related to exposing this AI pattern in JSM portals, so we can follow and vote on it.

The key requirement is to have a communication interface with AI inside the help experience, where the user can resolve issues first and only then create a ticket when needed, similar to what Atlassian does in its own support contact flow.

Thanks a lot in advance for any insights, patterns, or links you can share!

Best regards,
Mariano

[Rebekka Heilmann _viadee_ · Accepted solution]

Hi @Mariano Tallarico _Appricity_ 

1. The Atlassian Support portal certainly has some customizations/add ons that are not available via the regular JSM features.

2. With the new Service Collection, you can make Rovo Agents available in the Help Center or each Portal. It will pop up as *an optional* Chat window. Rovo can answer questions directly, provide knowledge base articles or raise a request on behalf of the customer. Raising requests is still quite limited and only works with Standard Jira fields.

Rovo cannot yet prefill a request form like the AI does on the Support site. It will either send a request directly or only provide knowledge.

There is probably a lot possible with Forge but I couldn't (yet) speak to that. Not my area of expertise.

3. Again, no expert on that. I am sure you could build something with forge, possibly levering the existing Rovo features. My colleagues only built a custom Chat interface so far that either provides self help or raises requests behind the scenes - so similar to Rovo but independent of the customer portal.

Hope this helps a little.

[Mariano Tallarico _Appricity_]

Thank you so much, Rebekka – your answer really helped clarify things.

Based on what you shared, I’m taking away the following:

• The Atlassian Support experience includes customizations that are not yet available as standard JSM Cloud features.
• Right now, the closest “official” approach is to use Rovo Agents in the Help Center / portals via Service Collections, letting them answer questions, suggest KB articles and, in simple scenarios, raise requests.
• To get closer to an “AI before ticket creation” pattern, it likely requires a more customized architecture (Forge + APIs / custom UI), possibly leveraging Rovo capabilities but outside of the standard portal form.

With that in mind, I’ll keep exploring two tracks in parallel:

1. Implementing Rovo in the Help Center to cover the most straightforward self-service use cases.
2. Investigating Forge and custom-built options to approximate this “pre-ticket AI layer” pattern for some of my more advanced customers.

If you or your team ever publish an example, talk, or article about the architecture of that custom chat you mentioned, I’d be very interested in reading it :)

Thanks again for taking the time to reply and share your experience!