Use the Prompt
Add Your Brief
Add your context below. It will be included with the prompt when you copy or download.
Create an interactive frontend prototype using the existing project stack and explicitly labeled demo data for [PROJECT] and [AUDIENCE]. The task is: AI Chat Workspace.
Required content and structure
Place conversation, source references, attachments and editable output in a split workspace.
Use [APPROVED_CONTENT] as the source of names, dates, figures, claims and destinations. List missing inputs before substituting anything. If a demonstration is needed, label its content as fictional. Do not invent customer quotes, rankings, partner logos or product capabilities.
Design direction
Use [BRAND_RULES] and the design system already selected in the project. If no system is supplied, start with this direction: warm white, ink text and restrained mint selected states; generous spacing and simple controls. Keep type, spacing and color roles consistent. Prioritize the task over decorative elements and avoid a generic collection of unrelated cards.
Content preparation
Define conversation records, attachments, source references and editable output. Supply examples of an empty thread, ongoing generation, interruption and failure.
Layout and behavior
Keep the conversation and working output distinct but connected. Show which source or attachment belongs to a response and make the output's editable state clear.
Task-specific acceptance check
- Show generating, interrupted, failed and empty states without inventing confidence scores.
- Interrupt generation and retry a failed response.
- Preserve typed drafts and earlier output, and do not display invented confidence percentages as a substitute for actual source information.
Implementation and review
Keep text editable. Use semantic structure, descriptive controls, visible keyboard focus and appropriate empty/error states. Check narrow and wide layouts. Respect reduced-motion preferences. Preserve any existing project framework and dependencies; do not silently replace the stack. Return the editable files and explain any incomplete behavior or unsupported inputs. Review this specific task before proposing extra features.
Reference attachment and reconstruction goal
Attach reference.webp from this prompt’s download bundle alongside this prompt in OpenDesign. Build an editable result with a similar composition and hierarchy; do not embed the whole reference image as the finished interface. Preserve approved project content and brand rules if they conflict with fictional sample labels. If the image is missing or unreadable, ask for it instead of claiming visual fidelity.
Reference composition and content
Use the attached reference.webp as the visual reference for AI Chat Workspace. Before building, identify the major regions from top to bottom, their relative widths and heights, alignment, spacing, typography hierarchy, surface colors, borders, imagery and repeated components. Preserve those relationships rather than substituting a generic template. Required task content: Place conversation, source references, attachments and editable output in a split workspace. Use the following task-specific layout guidance when translating the image into editable elements: Keep the conversation and working output distinct but connected. Show which source or attachment belongs to a response and make the output's editable state clear.
Visual direction from the example brief
warm white, ink text and restrained mint selected states; generous spacing and simple controls.
Interaction specification
The reference is a static concept: the behavior below is a requirement to implement, not proof that the picture already works. Copy the customized prompt into your working project. Start with the core task: Place conversation, source references, attachments and editable output in a split workspace. Show generating, interrupted, failed and empty states without inventing confidence scores. For each visible control, specify its trigger, affected state and feedback; implement local changes with consistent state across related views. Give search and filters real results, selection controls an explicit active state, and save actions visible pending/success/failure feedback where relevant. Do not add unrelated workflows. Use proper links for navigation, buttons for actions and accessible names for icon controls. Do not imply a backend operation succeeded when only a local demo changed.
Visual comparison and delivery
- Compare the result with the attached image at matching proportions. Review major region placement, relative image sizes, line breaks, spacing, typography hierarchy, contrast, active states and content density before polishing decorative details.
- Then test the narrow layout without squeezing desktop text into tiny labels.
- Keep logos, photos and icons as separate assets and all meaningful labels editable. List any unavailable fonts/assets and any intentional departures. Return the working files plus a concise account of implemented behavior and remaining integrations; do not promise a pixel-identical result from the prompt alone.
Download Prompt + Image
Includes the complete prompt and full-size reference image. Attach both in OpenDesign to guide the layout.
How to Write an AI Chat Workspace Design Prompt
Use this ai chat workspace design prompt to place conversation, source references, attachments and editable output in a split workspace.
The reference is an AI-generated concept. The instructions below describe what to build and test; the image does not demonstrate working behavior.
Keep Sources and Output in Context
An AI chat workspace needs room for conversation, references and editable output without losing their relationship.
Generating and interrupted output should be distinguishable from a completed response. Keep the conversation and working output distinct but connected. Show which source or attachment belongs to a response and make the output's editable state clear.
From the Prompt
Keep the conversation and working output distinct but connected.
Make It Your Own
Use a conversation pane beside editable output, with attached references available in context. Label generating, interrupted and failed states and preserve user edits when a response is retried.
Prepare the AI Chat Workspace Content
The right rail lists source files independently from the conversation. Supply document names and attachment types, and distinguish available sources from generated output.
Define conversation records, attachments, source references and editable output. Supply examples of an empty thread, ongoing generation, interruption and failure.
From the Prompt
Place conversation, source references, attachments and editable output in a split workspace.
Make It Your Own
For [PROJECT], use [APPROVED_CONTENT] as the source for this ai chat workspace. Define conversation records, attachments, source references and editable output. Supply examples of an empty thread, ongoing generation, interruption and failure. Identify missing inputs before replacing them with demonstration content.
Interrupt a Response
The conversation shows generating, interrupted and error examples with Continue and Retry controls. Define separate transitions for those states; the screenshot is a state inventory, not evidence of a running model.
The partial result should remain identifiable and recoverable. Retrying should not silently replace user-written changes or invent a confidence score for the answer. Interrupt generation and retry a failed response. Preserve typed drafts and earlier output, and do not display invented confidence percentages as a substitute for actual source information.
From the Prompt
Show generating, interrupted, failed and empty states without inventing confidence scores.
Make It Your Own
Review the ai chat workspace against this requirement: Show generating, interrupted, failed and empty states without inventing confidence scores. Interrupt generation and retry a failed response. Preserve typed drafts and earlier output, and do not display invented confidence percentages as a substitute for actual source information. Preserve approved content and revise only the parts that fail this check.
Review Your AI Chat Workspace Result
A convincing screen is not a working backend. Test validation, permissions, persistence and failure behavior separately before connecting real data.
- Interrupt generation and retry a failed response. Preserve typed drafts and earlier output, and do not display invented confidence percentages as a substitute for actual source information.
Follow-Up Prompt
Use this after the first result. It narrows the revision to the most important part of this task.
Refine the current ai chat workspace. Interrupt generation and retry a failed response. Preserve typed drafts and earlier output, and do not display invented confidence percentages as a substitute for actual source information. Preserve the approved content and visual direction; change only the parts needed to resolve these issues.
