UX Briefing from a Ticket: Personas, Scope, Open Questions and Risks

Goal

Start a design from a Jira ticket or story with a structured briefing, before any frame exists.

Before you start

• A ticket or user story of at least 200 characters.
• An AI provider set up in AI Settings. The UX Feedback view asks for a provider that can read images, even though a briefing itself is text only.

Steps

1. Open UX Feedback with nothing selected Click the eye button (UX Feedback) in the sidebar and clear your selection in Figma. The prompt box then reads 'Select elements to analyze — or paste a ticket for a UX briefing'.
2. Paste the ticket Paste the whole ticket into the prompt box: its title, description and acceptance criteria. Contently takes it for a ticket when it is at least 200 characters long and shows two or more of these signs: 'As a … I want', 'acceptance criteria', Given … when … then, story points, a line that starts 'Assignee:', 'Reporter:', 'Epic Link:' or 'Sprint:', definition of done, a Jira heading (h1. to h6.), or a key such as SHOP-318. One sign is enough when the text is longer than 500 characters.

   

3. Send it Press Enter or the send button. The briefing runs on your AI provider and counts as one AI action. If anything is selected in Figma, the same text is used as the prompt of a normal review instead, so clear the selection first.
4. Read the briefing The report opens with tabs for In One Line, Personas & Needs (the job to be done, persona cards and key needs), Scope, Key Flows, Acceptance Criteria as design implications, Open Questions, and Risks & Constraints. A section the ticket gives nothing for is left out. A briefing has no scores.

   

5. Render it or find it again The briefing is saved to your Feedback history when it finishes, titled with the start of the ticket and with no score, so you can open it again. Back in the report returns to the first screen, where the labelled Render to Figma button is; inside a section the button is the icon next to Back. It draws the briefing as a frame in the middle of your view, or beside the selected layer if one is selected, and offers the same render options as a review.

Tips

• Add the product's audience and tone to Document Context (AI Settings → Context); the briefing uses it.
• The ticket text is kept in your prompt history in the plugin. Remove it from the history if the ticket is confidential.
• The briefing is told not to invent a persona: when the ticket does not say who it is for, that shows up under Open Questions.

Result

You have a briefing to design from, and a list of questions to ask the team before you start.

If something goes wrong

• Nothing happens: the spinner stops and the text stays in the box.

  The text was not taken for a ticket: it is under 200 characters or shows too few of the signs above. Paste the whole ticket with its headings and acceptance criteria, or select a frame to get a normal review.

• My ticket was treated as a review prompt.

  Something is still selected in Figma. Clear the selection and send it again.

• The button says to try the Playground, or a note says vision feedback needs a provider with vision capability.

  Choose a provider that can read images in AI Settings → API Key, then run Test & Detect Capabilities.

• A message says today's free AI uses are used up.

  The free plan has 12 AI actions a day, shared by every AI feature. They come back the next day; a paid upgrade removes the limit.

Related guides

• AI ASSISTANT & VISION
  UX Feedback Reports: Read the Score, Work Through the Tasks, Render to Figma Understand the Overall Score, the lens tabs with their tickable tasks, the history of earlier reviews and the Render to Figma options.
  →
• AI ASSISTANT & VISION
  AI Assistant, Vision Reviews & Layer Renaming Use Contently's AI for multimodal design feedback, layer renaming from image contents, copy brainstorming, and component-aware layout builds.
  →
• TROUBLESHOOTING & LOCAL AI
  Local AI Setup (LM Studio, Ollama), CORS & Troubleshooting Run open-weights vision and text models on your own hardware via LM Studio or Ollama, so AI requests go to your own machine instead of a cloud AI provider.
  →