Overview
Hatz AI offers two complementary features that customize your AI interactions: Personalization and Memory. While they work together to improve your experience, they serve distinct purposes.
Personalization is a static profile you manually configure that controls how the AI responds to you (style, tone, format). Memory is a dynamic knowledge base the AI builds automatically about you over time that controls what the AI knows about you (facts, context, history).
Who this is for
This feature is available to all Hatz users who want more personalized AI interactions. Both features apply at the individual user level within a workspace.
Summary
Personalization controls how the AI responds—style, tone, length, and behavior. It's a static profile you manually create and update.
Memory controls what the AI knows about you—facts, context, and history. It's a dynamic knowledge base the AI builds automatically during conversations.
Together, they create a consistent, contextually aware AI experience tailored to your unique needs.
What is Personalization?
Personalization is a static profile you configure manually. You fill it out once during onboarding and can update it anytime in Settings > AI Preferences.
What Personalization stores:
Your name, timezone, and job description
Response length preference (Concise, Balanced, or Detailed)
Tone preference (Neutral, Formal, or Friendly)
Emoji usage preference (On, Off, or Mirror User)
Custom freeform instructions (e.g., "always summarize key actions first")
How Personalization works:
Your preferences are compiled into a pre-generated system prompt
This prompt is injected at the top of the system message for every chat
It shapes how the AI responds—controlling tone, length, and formatting style
User control:
You explicitly write and save these preferences. Nothing changes unless you manually update and save new settings in AI Preferences.
What is Memory?
Memory is a dynamic, auto-growing knowledge base about you that the AI builds over time during conversations.
What Memory stores:
Individual fact and summary items (e.g., "User enjoys spicy food", "User travels to New York frequently")
Each memory has an importance score and a vector embedding for semantic search
Memories are stored and recalled based on relevance to the current conversation
How Memory works:
During conversations, the AI can automatically save, forget, search, and list memories using built-in tools
Before each AI response, the system predicts which memories are relevant (based on prior message context), fetches them, and injects them as a memory block in the system message
User control:
Memory is mostly automatic—the AI decides what to save based on conversation context. However, you can review and manually manage your full memory list at any time.
The core distinction
| Personalization | Memory |
Nature | Static profile you write | Dynamic facts the AI learns |
Controls | HOW the AI responds (style/tone) | WHAT the AI knows about you |
Updated by | You (manually) | AI (automatically, during chats) |
Applied | Every chat, always | Only when relevant to the current chat |
Scope | Formatting, tone, instructions | Facts, context, history |
Expected results
With Personalization enabled:
AI responses will consistently match your configured style, tone, and length preferences
The AI will reference your personal details (name, job description, timezone) in responses
Custom instructions will be applied across all conversations and models
With Memory enabled:
The AI will recall relevant facts and context that created a memory previous conversations
Responses will become more contextually aware over time as the AI learns more about you
You'll see continuity across conversations without repeating information
With both enabled:
The AI will respond in your preferred style (Personalization) while incorporating relevant facts it has learned about you (Memory)
You get both consistent formatting and intelligent context awareness
Troubleshooting
Personalization changes aren't taking effect
Ensure you clicked Save Settings after making changes
Refresh the page and start a new conversation to see updated preferences
Verify that the Personalization toggle is enabled in AI Behavior
Memory isn't recalling information
Confirm that the Memory toggle is enabled in AI Behavior settings
Check your Memory list to verify the information was actually saved
Note that Memory is recalled selectively—only memories relevant to the current conversation are injected
AI is not saving new memories automatically
Ensure Memory is enabled in AI Preferences
The AI prioritizes important information—not every detail will be saved
You can manually add memories in Settings > Memory if needed
Unclear what to include in Custom instructions
Click See examples in the Custom instructions field for guidance on effective instructions
Focus on behavioral preferences rather than facts (facts belong in Memory)
Limitations
Personalization limitations:
Personalization shapes how the AI responds but does not change the underlying capabilities of the models
Custom instructions are limited to 2,000 characters
Changes only affect new AI interactions; existing conversation history is not retroactively modified
Response length, Tone, and Emoji options are limited to the preset choices provided in the interface
Personalization must be enabled in AI Behavior settings to function
Memory limitations:
Memory must be enabled in AI Behavior settings to function
The AI determines what information is important enough to save automatically—not all conversational details are stored
Memories are recalled based on relevance to the current conversation, so not all memories appear in every chat
Manual management of memories (editing, deleting, merging) requires accessing Settings > Memory
Memory does not retroactively modify previous conversation history; it only influences future responses
The semantic search system may not always retrieve the most relevant memories for highly ambiguous queries
Combined limitations:
Neither feature modifies the capabilities or knowledge cutoff dates of the underlying language models
Both features require explicit user enablement via toggles in AI Preferences
Organization-level policies and model availability restrictions set by your admin take priority over personal preferences