Leveraging AI to minimize incomplete requests and agent effort

Question

Hi everyone! 👋

I have a question about Jira Service Management (JSM), Rovo, and AI capabilities.

One of my clients has recently migrated to JSM Cloud and is exploring how AI features could help improve their service desk operations. Overall, they’re happy with Cloud, but they want to leverage AI to make life easier for their agents.

They handle a high volume of customers and daily ticket creation. JSM is used for both ITSM and non-IT service management. Agents act as first-line support and frequently forward tickets to developers and BA/PM teams.

A major challenge is incomplete or poorly structured requests. Customers often provide insufficient or irrelevant information, which leads to a lot of back-and-forth between customers, agents, and delivery teams, resulting in significant time loss and delays in request resolution.

Even though customers are provided with clear instructions on how to raise requests, as many of you have probably experienced, those instructions are often ignored.

The idea

The client is wondering whether AI could act as a “first checkpoint” before a request is submitted or reaches an agent.

Here’s a simplified example:

* John works in the HR department and submits a request for a new feature in an internal app related to budgets and payouts
* His request should ideally explain what is needed, why, how it should work, and the business process behind it
* Historically, agents often have to ask John follow-up questions because he struggles to clearly describe processes
* Additionally, HR requests often miss legal considerations, resulting in further delays and back-and-forth with the Legal team.

What the client imagines AI could do

At the moment of ticket creation, AI would ideally be able to:

* Recognize that John typically provides weak process descriptions and prompt him to add more details (or help enrich the request)
* Recognize patterns from previous HR requests and prompt for legal-related information when it’s missing
* Analyze historical tickets to determine whether the request contains enough information before allowing it to be submitted.

My questions

* Is something like this achievable with AI / Rovo in JSM?
* Can AI learn from historical requests and patterns?
* Can AI be used during request creation on the portal, or does the request need to be created first before AI can process it?
* Has anyone implemented something similar, or found good workarounds?

Any tips, experiences, or insights would be highly appreciated. Thanks in advance! 🙌

Marc -Devoteam- · Accepted Answer

Hi @Nina Dragušica

Welcome to the community.

Your first option would be to dive into the virtual agent for JSM.

https://www.atlassian.com/software/jira/service-management/product-guide/tips-and-tricks/virtual-agent#virtual-service-agent-101

Here user can be guided in creating a request.

Second, make sure on requests fields for specific relevant information are required.

Make sure to have a Confluence KB attached with articles to help user solver their own issue or to provide steps on how a request should be provided

Dr Valeri Colon _Connect Centric_ · Accepted Answer

@Nina Dragušica welcome to the community - Parts of this are achievable today, with some limits. Rovo in JSM can analyze requests after submission to summarize, flag missing info, and suggest follow-up questions, but it can’t yet block or validate a request before creation on the portal. It also doesn’t “learn” autonomously from history, but you can guide behavior using request types, required fields, forms, and automation + Rovo prompts (for example, auto-comment asking for legal or process details based on keywords). The strongest pattern today is combining JSM forms + automation + Rovo post-creation checks rather than true pre-submission AI gating.