
Written by asad
Last updated 2 months ago
Smart Memory is currently in Beta and available for all users, rolling out gradually across the platform.
The Problem Smart Memory Solves
Until recently, MemoryPlugin operated as a simple list. You would add new memories to the list, and the entire list would be shared with the AI when required. While this is useful, it has significant limitations:Irrelevant Information
Dumping the entire memory list feeds tons of irrelevant information to the AI, potentially reducing performance by confusing the model
Context Window Bloat
Not every conversation needs full details of every memory. As you add more memories, they take up more space, leaving less room for your actual work
Higher Costs
All API-based tools charge by the token, so more information in the context window means higher costs and faster rate limiting
What is Smart Memory?
Smart Memory automatically analyzes and organizes your memories into intelligent categories with AI-generated summaries.How It Works
Smart Memory operates at the Memory Bucket level:- Analysis: AI analyzes all memories within a bucket to identify themes and patterns
- Categorization: Groups related memories into meaningful categories like “Work Projects”, “Learning Goals”, “Health & Wellness”
- Summarization: Creates concise summaries for each category
- Selective Loading: AI can load specific categories relevant to the conversation context
Instead of loading all memories, the AI can focus on just the categories relevant to your conversation, making interactions more efficient and contextual.
Benefits
Better Context Efficiency
Smart Memory improves AI interactions by:- Loading only relevant memory categories instead of everything
- Providing focused context that matches your conversation topic
- Preserving context window space for actual conversation
Improved Organization
- Automatic categorization: AI identifies themes and groups related memories
- Clear summaries: Each category gets a concise overview of its contents
- Easy filtering: Focus on specific categories in the dashboard
Enhanced Scalability
- Handle more memories: Categories keep large collections organized and accessible
- Maintain performance: Selective loading prevents information overload
- Better relevance: AI gets context that actually matters for the conversation
How Smart Memory Categories Work
Category Creation
Smart Memory analyzes your memories to create categories:Requirements and Limits
- Minimum: 30 memories required in a bucket for categorization
- Maximum: Works with buckets up to 600,000 tokens or 2,000 memories
- Manual Process: Categories must be created by clicking the dashboard button
Using Smart Memory
Getting Started
Smart Memory requires manual activation in your dashboard:- Add 30+ memories to any bucket
- Click “Create Smart Categories” button in the dashboard
- Wait for processing (usually takes a few minutes)
- View your organized categories with summaries and additional context
In the Dashboard
Once categories are created, you can:- Browse categories: Each shows a summary and additional information
- Filter by category: Focus on specific themes
- Refresh categories: Click the refresh button to update organization
- View memory counts: See how many memories are in each category
Category Information
Each category provides multiple layers of information:- Category Name: AI-generated theme (e.g., “Work Projects”, “Health Goals”)
- Summary: High-level overview of what memories in this category contain
- Additional Information: Deeper context about the category’s contents, including:
- Key topics and themes within the category
- Specific details that distinguish this category from others
- Context that helps AI assistants understand when this category is relevant
- More granular information about the types of memories included
- Memory Count: Number of memories grouped in this category
Availability
Smart Memory is currently in Beta and rolling out gradually to all users.With AI Integrations
Once you have Smart Memory categories set up:- Browser Extension: Categories are available when using AI platforms like ChatGPT, Claude, etc.
- MCP Server: Smart Memory integration is being developed for MCP-compatible clients
- Custom GPT: Smart Memory integration is planned for future development
- API: Smart Memory endpoints are currently internal-only
Use Cases
Smart Memory is particularly valuable for:Power Users
- Large memory collections (hundreds or thousands of memories)
- Multiple complex projects with extensive documentation
- Long-term AI assistance relationships with deep context
Professional Applications
- Customer service with extensive customer history
- Consulting work with detailed client information
- Research projects with vast amounts of collected information
Cost-Conscious Users
- High-volume API usage where token costs matter
- Rate-limited environments where efficiency is crucial
- Performance-sensitive applications
Example Categories
Smart Memory might organize your memories into categories like:Work Projects
Meeting notes, project updates, deadlines, team communications
Health & Wellness
Exercise routines, nutrition goals, medical appointments, wellness tracking
Learning & Development
Course notes, tutorials, skill development, book summaries
Personal Goals
Life goals, habit tracking, personal reflections, milestone planning
Comparison: Traditional vs Smart Memory
Aspect | Traditional Memory | Smart Memory |
---|---|---|
Organization | Chronological list of all memories | Themed categories with summaries |
Context Loading | All memories loaded at once | Relevant categories loaded selectively |
Scalability | Gets unwieldy with many memories | Stays organized regardless of size |
AI Relevance | AI sees everything, relevant or not | AI gets focused, contextual information |
Dashboard View | Simple memory list | Organized by categories with filtering |