It’s now remarkably easy to produce deliverables, as AI can generate wireframes, diagrams, HTML prototypes, and polished mockups in minutes. This rapid creation is reminiscent of when product managers and founders would use any available tool, like PowerPoint, to visualize an idea. Increasingly, I receive high-fidelity prototypes before seeing a simple brief with core product thinking or requirements.
I’m not here to sound like an old-school designer; AI can absolutely help in the early stages of design. Most of the time, you can tell if it’s really adding value or just creating noise. If it’s not helping, this post offers a few ideas that may be useful.
Using the right tool at the right time
With AI, just because you can doesn’t always mean you should.
The challenge is that AI-generated deliverables often introduce detailed UI before the underlying concepts, relationships, and mental model are established. Teams end up debating screens before aligning on what the system is, how it works, or the right user experience. This makes it harder to extract the core intent, as AI produces more noise than clarity.
While prototypes have their place, it’s usually easier to start with organized text and lightweight models than with fully formed artifacts.
It’s natural to want to visualize ideas, as screens, diagrams, and prototypes help us communicate and explore. But these can sometimes create the illusion of clarity before we’ve truly defined the essentials.
To make ideas easier to discuss and collaborate on, I’ve developed simple methods you can use, alone, with stakeholders, or with AI, to clarify product thinking before design.
Use AI and your own thinking to nail the fundamentals first
The early design process is about working through foundational questions together. Before designing interfaces and along the way, we need to clarify the core concepts, relationships, information architecture, and user goals that shape the product.
Focusing on concepts and relationships first makes collaboration easier for everyone, shifting the emphasis from screens and layouts to the ideas that matter most.
Visualizing ideas is still valuable, and the goal isn’t to avoid visual artifacts but to use lighter-weight representations to communicate ideas without jumping ahead to finished design.
With that in mind, I’ve put together techniques to communicate concepts, relationships, structure, and intent before moving into wireframes or visual design. Each serves a different purpose, with example AI prompts included.
10 lightweight artifacts for clarifying product thinking before design
To demonstrate each technique, I’ll use a fictional team wiki product. The aim isn’t to design the product itself, but to show how different artifacts can clarify concepts, relationships, structure, and intent before moving into design.
1 - Concept statement
- What is it: A brief summary that defines what the product, feature, or concept is and its purpose.
- Question it answers: What is this thing?
Example
A team wiki helps teams create, organize, and find shared knowledge in one place.Prompt
Define the core concepts in this system. For each concept, provide a concise definition in one or two sentences. Focus on what the thing is, not how it is implemented.
2 - Entity model (object model)
- What is it: A list of the main objects in the system, typically the key nouns users interact with or manage.
- Question it answers: What are the elements in the system?
Example
Workspace
Collection
Page
Author
CommentPrompt
Identify the primary entities in this system. Focus on the core nouns users need to understand and manage. Avoid UI concepts, pages, and implementation details.
3 - Relationship diagram
- What is it: A visual diagram showing how system entities are connected.
- Question it answers: How are the entities related?
Example
Workspace
└─ Collection
└─ Page
├─ Author
└─ CommentPrompt
Create a simple relationship diagram showing how the primary entities relate to one another. Use only boxes, indentation, and arrows. Do not design the UI.
4 - Cardinality model
- What is it: Describes how many of one entity relate to another (e.g., one-to-one, one-to-many).
- Question it answers: How many of each thing can exist?
Example
A Workspace can have many Collections
A Collection can have many Pages
A Page belongs to one Collection
A Page has one Author
A Page can have many CommentsPrompt
Define the cardinality relationships between the core entities. Describe ownership and one-to-many relationships using plain language.
5 - Information architecture (IA)
- What is it: A high-level overview of how information, pages, and concepts are structured in a product.
- Question it answers: How is the product organized?
Example
Workspace
├─ Collections
├─ Pages
├─ Search
└─ SettingsPrompt
Create a high-level information architecture showing navigation, page hierarchy, and user flow. Focus on structure, not interface design.
6 - Sitemap
- What is it: A hierarchical outline of pages showing product structure and user navigation paths.
- Question it answers: What pages exist?
Example
Home
Collections
└─ Collection detail
└─ Page detail
Pages
└─ Page detail
Search
SettingsPrompt
Create a sitemap that shows all pages and subpages. Use a simple hierarchical structure. Do not include UI components or layouts.
7 - Content inventory
- What is it: A complete list of all content, data, and actions found on a page.
- Question it answers: What information appears on each page?
Example
Page detail
├─ Page title
├─ Page content
├─ Author
├─ Last updated
├─ Collection
├─ Comments
├─ Related pages
└─ Edit historyPrompt
Create a content inventory for each page. List all information, data, and actions that should appear on the page. Do not describe layout or visual design.
8 - Page blueprint
- What is it: A basic wireframe showing a page’s main sections and content groups.
- Question it answers: What sections are on the page?
Example
┌──────────────────────────────┐
│ Page detail │
├──────────────────────────────┤
│ Page title │
│ Metadata │
│ │
│ Page content │
│ │
│ Related pages │
│ Comments │
│ Edit history │
└──────────────────────────────┘Prompt
Create a low-fidelity page blueprint. Organize content into logical sections and containers. Do not design components, styling, or visual treatments.
9 - Feature inventory
- What is it: A list of all actions, capabilities, and features available in the product.
- Question it answers: What can users do?
Example
Workspace
• Create collection
• Create page
• Edit page
• Comment on page
• Search pages
• Move page between collections
• View edit historyPrompt
List the actions and capabilities available for each entity. Focus on user actions rather than interface controls.
10 - User tasks/jobs
- What is it: Describes the tasks or goals users want to achieve with the product.
- Question it answers: Why does this product exist?
Example
• Create shared documentation
• Find information quickly
• Organize knowledge by topic
• Keep content up to date
• Discuss or clarify information
• Understand what changed over timePrompt
For each page or entity, identify the primary user jobs and goals. Focus on what users are trying to accomplish, not the features used to accomplish them.
These techniques are not new.
Designers have used versions of them for decades.
What’s changed is how easy it is to jump straight to deliverables. AI can now generate wireframes, prototypes, diagrams, and polished mockups in minutes, making it tempting to skip the thinking that usually comes first.
The goal isn’t to replace prototypes or design artifacts, but to use the right tool at the right time. Sometimes a one-sentence concept statement is more valuable than a prototype; sometimes a simple relationship diagram unlocks more understanding than a wireframe.
If these techniques help you have clearer conversations, write better briefs, or align teams faster, please let me know or leave a comment!
