Dashboard Gadget (AI Report)

The dashboard gadget (AI report, standing question, dashboard widget) answers one saved question and shows the answer on a Jira dashboard — for example a sprint summary, a status overview of project ABC or a list of open risks. This page explains how to add and configure the gadget, how the answer is refreshed, who sees it and which limits apply.

What this does

You save a question once; the gadget shows the AI's answer to it on the dashboard. The answer is generated on demand with ↻ Refresh and stays visible until someone refreshes it again. Optionally the gadget uses one of your agents and a specific model. In the gadget directory the gadget is listed as AnyLLM Chat with the description Shows the AI answer to a saved question — e.g. sprint status or open risks.; administrators see it in the module list as Dashboard gadget (AI report).

Who can do this

• Set up: anyone who may edit the dashboard.
• Refresh and read: anyone who can see the dashboard. A refresh counts against the personal hourly limit of the person who clicks ↻ Refresh.

Before you start

• The app has been set up by your administration and the subscription is active.
• The module Dashboard gadget (AI report) is enabled. If it is switched off, the gadget — including its edit dialog — shows only This feature has been disabled by your administrator.
• Think about the audience: the answer is generated with the Jira and Confluence permissions of the person who refreshes it and is visible to everyone who can see the dashboard.

Step by step: adding and configuring the gadget

1. Open the dashboard and add a gadget from Jira's gadget directory.
2. Choose AnyLLM Chat (description: Shows the AI answer to a saved question — e.g. sprint status or open risks.).
3. In the gadget's edit form fill in Question for the AI (answered on the dashboard) — required. The placeholder suggests: e.g. Summarize the status of open issues in project ABC — risks first.
4. Optionally choose an Agent (optional) — the list shows your available agents as "emoji name"; No agent is the default.
5. Optionally choose a Model (optional) — the list shows models as "connection name · model"; Default uses the site's default model.
6. Read the note Note: the answer is generated with the Jira/Confluence permissions of the person who clicks "Refresh", and is visible to everyone who can see this dashboard.
7. Click Save.

⚠️ Warning: Only the first 2,000 characters of the question are saved; longer text is cut off silently when you click Save. If you save with an empty field, the gadget shows Please enter a question.

Step by step: refreshing the answer

1. In the gadget, click ↻ Refresh (announced to screen readers as Regenerate answer). Before the first refresh the gadget reads No answer yet — click "↻ Refresh".
2. Wait for the answer. It streams live into the gadget; until the first text arrives you see Generating answer …. The previous answer stays visible and is replaced only once the new one is complete.

Expected result

The gadget shows the question at the top (hover over a shortened question to see the full text), the answer as formatted text and below it the line As of: followed by the date and time (in the format of your interface language) and the model that produced the answer, e.g. As of: … · gpt-4o. Images in the answer are not loaded automatically; they appear as the link view external image, which opens in a new tab.

How the gadget behaves

• Visibility: the answer is visible to everyone who can see the dashboard — even if the question touches content that only the refreshing person may see. Phrase the question accordingly.
• Permissions: the AI reads Jira and Confluence with the permissions of the person who clicks ↻ Refresh; two people can therefore get different answers to the same question.
• Persistence: the answer is stored with the gadget, not in anyone's chat history. It is shown to every viewer each time the dashboard loads, survives page reloads, has no expiry and is not deleted when a user account is closed. It is replaced only by the next ↻ Refresh.
• No automatic refresh: nothing happens until someone clicks ↻ Refresh.
• Rate limit: every refresh counts as one request against the personal hourly limit of the person refreshing — the same counter as in the chat, across Jira and Confluence.
• Language: the gadget follows the interface language set by the administration; with Follow each user’s Atlassian language it follows your Atlassian language.

Common problems

More messages: Troubleshooting.

Related topics

• Getting Started
• AnyLLM in Jira
• Creating Your Own Agents and Skills — agents you can choose in the gadget
• Permissions, Roles and Data Visibility
• Modules: Switching Surfaces On and Off
• General Settings: Language, Default Model, AI Access, Rate Limit, Site Knowledge — default model and rate limit
• Troubleshooting · FAQ