> ## Documentation Index
> Fetch the complete documentation index at: https://docs.figr.design/llms.txt
> Use this file to discover all available pages before exploring further.
# Memories
> Figr remembers your product, your patterns, and your preferences. Every conversation makes it smarter.
***
## How Memory Works
Figr listens to your conversations and extracts meaningful context. When you describe how your navigation works, explain a design decision, or share your content style, Figr stores it.
Memory builds passively. You don't need to do anything special. Just work with Figr, and it learns.
* Describing your product structure
* Explaining design decisions and rationale
* Sharing user flow patterns
* Discussing content and copy style
* Stating preferences and constraints
## Viewing Memories
### Open the Memory Panel
1. Click **Memories** in the left sidebar
2. Browse all stored memories
3. Use the search bar to find specific context
***
## Editing Memories
Memories aren't locked. Update them as your product evolves.
### Edit a Memory
1. Open the Memories panel in the sidebar
2. Find the memory you want to update
3. Double-click to open edit mode
4. Make your changes
5. Press **Save** (⌘↵) or click Save button
### Delete a Memory
When context becomes outdated or incorrect:
1. Double-click the memory to open it
2. Click **Delete** (bottom-left, red text)
3. Confirm deletion
## Why Memory Matters
Every other AI tool has amnesia. You explain your product, describe your patterns, share your constraints. Get a decent result. Tomorrow? Start over. Figr remembers everything.
Your design patterns. Your user flows. Your content style. That decision you made three months ago about button placement. The way you like tables structured. All of it compounds into deeper product understanding over time.
| **Traditional AI Tools** | **Figr's Memory** |
| -------------------------------------------------------- | ---------------------------------------------------------- |
| Forget everything between sessions | Learns from every conversation automatically |
| Require re-explaining context every time | Builds deeper understanding over time |
| Generate generic outputs that miss your product's nuance | Generates designs that match your existing product |
| Can't learn from past decisions or feedback | Remembers decisions, patterns, and preferences permanently |
## Memory Categories
The foundational details about what you're building.
```text theme={null}
Examples:
- "Dashboard is the entry point after login,
shows key metrics and recent activity"
- "User types include Admin, Manager, and Viewer
with different permission levels"
- "Primary navigation is a left sidebar that
collapses on mobile"
```
How your product looks and behaves visually.
```text theme={null}
Examples:
- "Primary CTA uses solid blue/purple button
with icon, positioned top-right"
- "Delete actions use red color to signal
destructive intent"
- "Disabled states shown in gray text"
- "Cards use subtle shadow with 8px radius"
```
How users move through your product.
```text theme={null}
Examples:
- "Manage Members screen is entry point for
organization-level member and seat management"
- "Reducing seats flow: click 'Manage plan seats'
→ view allocation → select reduction → confirm
→ handle member removal if necessary"
- "Onboarding uses progressive disclosure across
4 steps with skip option on non-critical steps"
```
How you prefer to work with Figr.
```text theme={null}
Examples:
- "Prefers seeing 2-3 variations before committing
to a direction"
- "Wants edge cases called out proactively"
- "Likes detailed rationale with design decisions"
- "Prefers mobile-first approach for new features"
```
Your voice, tone, and copy conventions.
```text theme={null}
Examples:
- "Error messages are friendly, never blame the user"
- "CTAs use action verbs: 'Create project' not
'New project'"
- "Empty states include illustration and single
clear action"
- "Microcopy is concise, max 2 lines"
```
## Adding Memories Manually
In any conversation, explicitly ask Figr to store something.
```text theme={null}
Example prompts:
"Remember that we always use inline validation
for forms, never on-submit validation"
"Remember that our primary user persona is a
marketing manager at a mid-size B2B company"
"Remember that modals are only used for
destructive actions, never for forms"
"Remember that our brand voice is friendly
but professional, never casual"
```
Vague memories aren't useful. Specific ones are.
| **Less useful** | **More useful** |
| --------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| "Remember our button style" | "Remember that primary buttons are solid with #6366f1 background, secondary buttons are outline style, and destructive buttons use red background" |
| "Remember how forms work" | "Remember that forms use inline validation on blur, error messages appear below the field in red, and submit button stays disabled until all required fields are valid" |
***
## Memory in Action
| **Your prompt** | **Without Memory** | **With Memory** |
| ------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| "Design a settings page" | A settings page that looks like every other settings page. Wrong colors. Wrong patterns. Wrong component styles. You spend an hour fixing it. | A settings page that: - Uses your left sidebar navigation pattern - Follows your form layout conventions - Matches your button hierarchy (primary/secondary) - Includes your error state styling - Uses your typography and spacing system |