Getting Started With AI Assist
AI Assist generates layout suggestions, content drafts, and interaction patterns based on your project context. Instead of starting from a blank canvas, you describe what you need and AI Assist produces starting points you can accept, modify, or reject.
This guide walks through enabling AI Assist, configuring it for your project, and developing a workflow that uses AI suggestions effectively without losing design control.
What AI Assist solves
Building prototypes from scratch is slow when the goal is validation speed. Teams spend time on layout decisions, placeholder content, and repetitive interaction wiring that could be automated. AI Assist handles the mechanical work so the team can focus on the decisions that actually need human judgment.
AI Assist is not a replacement for design thinking. It produces suggestions that are good starting points but rarely perfect. The value comes from accelerating the first draft, not from accepting every suggestion without review. Teams that treat AI Assist as a collaborator rather than an oracle get the most benefit.
The alternative is manual construction of every screen, which is fine for final-fidelity work but wasteful during early validation when prototypes are disposable.
Getting started with AI Assist
- Open your workspace settings and enable AI Assist under the AI Features section. Workspace admins can enable it for the entire workspace or allow project-level opt-in.
- In any project, open a screen and use the AI Assist panel (right sidebar) to describe what you want. Be specific about the purpose, audience, and content type.
- Review the generated suggestion in the preview pane before inserting it. You can request variations by refining your description or adding constraints.
- Accept the suggestion to add it to your canvas, then customize it using standard editing tools. Overrides you make are preserved even if you regenerate the base suggestion later.
- Configure project-level AI preferences to set default constraints like design system adherence, content tone, and layout style. These preferences apply to all AI Assist requests in the project.
- Share effective prompts with your team by saving them to the workspace prompt library. This reduces duplication and improves consistency across projects.
First-time setup mistakes
- Enabling AI Assist without setting project-level preferences first. Without constraints, suggestions may not match your design system or content standards.
- Writing vague prompts like "make a dashboard" instead of specifying the data types, user role, and key actions the dashboard should support.
- Accepting AI suggestions without reviewing them, then discovering inconsistencies during stakeholder review. Always inspect generated content before sharing.
- Expecting AI Assist to understand product context it has not been given. Provide relevant background in your prompt or project description for better results.
- Using AI Assist for final-fidelity work where pixel precision matters. It is optimized for early-stage ideation and rapid prototyping, not production-ready design.
Tracking AI Assist adoption
- Suggestion acceptance rate: The percentage of AI suggestions your team uses versus discards. A rate between forty and seventy percent typically indicates good prompt quality and appropriate expectations.
- Time to first screen: How long it takes to create the first interactive screen in a new project. Teams using AI Assist effectively should see this decrease.
- Prompt refinement rate: How often users modify their prompt before accepting a suggestion. High refinement rates suggest prompts need better defaults or templates.
- Team coverage: The percentage of team members actively using AI Assist. Low coverage may indicate training gaps or trust issues that need addressing.
When AI Assist helps most
- During kickoff workshops when the team needs to rapidly explore multiple directions before committing to one.
- When building prototypes for user testing where speed matters more than pixel perfection.
- For generating variations of existing screens to test different layouts, content approaches, or interaction patterns.
- When onboarding new team members who can use AI Assist to learn the design patterns and conventions used in existing projects.
Key concepts
- AI Assist: The built-in AI capability that helps teams generate, refine, and iterate on prototype elements faster. It suggests layouts, content, and interaction patterns based on your project context.
- Prompt context: The information AI Assist uses to generate relevant suggestions, including your current prototype state, project goals, and team preferences.
- Suggestion acceptance rate: How often team members use AI-generated suggestions versus modifying or rejecting them. A useful proxy for whether the AI is providing relevant help.
FAQ
- Do I need AI experience to use AI Assist? No. AI Assist is designed for product and design professionals. It generates suggestions based on your prototype context, and you accept, modify, or reject each one.
- Can I control what AI Assist suggests? Yes. You can set project-level preferences, provide prompt context, and configure which types of suggestions appear.
- Does AI Assist work on existing prototypes? Yes. Open any existing project and AI Assist will analyze the current state to provide relevant suggestions.
Next steps
Begin with one repetitive task in your prototype workflow and configure AI Assist to handle it. Evaluate the output quality over five iterations before expanding to additional tasks. Use the FAQ above to troubleshoot common setup issues.