Skip to main content
Personal memory is your “user profile” built by Profy. As you converse with different Experts, the system automatically extracts your preferences, habits, and commonly used tools, forming memory entries that are shared across all Experts. These memories help every Expert understand you better.

What Is Memory

Personal memory consists of cognitive fragments naturally accumulated as you use Profy. For example:
  • “User prefers TypeScript”
  • “User’s go-to development framework is Next.js”
  • “User wants code comments written in Chinese”
  • “User is working on a SaaS project with Hono as the backend”
  • “User dislikes over-engineering and prefers a clean, straightforward code style”
These memories come from your everyday conversations—no manual input required.

How Memory Is Created

1

Conversation

You chat with any Expert—discussing requirements, writing code, analyzing data…
2

Intelligent Extraction

After the conversation, the system analyzes the exchange in the background and extracts valuable information. This process is completely transparent and does not interrupt your experience.
3

Quality Filtering

Not all information becomes a memory. The system applies three criteria:
  • Specificity: Only concrete, clear preferences are kept—not vague statements
  • Durability: Only long-term information is retained—not one-off context
  • Non-obviousness: Only insights that cannot be inferred from common sense
4

Cross-Expert Sharing

Extracted memories belong to you (the user), not to any particular Expert. All Experts can access these memories when chatting with you.
You can also explicitly create memory entries by telling an Expert “remember this” during a conversation. For example, tell an Expert “Remember: my project uses PostgreSQL 16.”

Managing Memory

You have full control over your memory entries. View and manage all memories under “Settings → Memory.”

Viewing the Memory List

The memory management page displays all your memory entries. Each entry includes:
  • Content: The specific information extracted
  • Source: Which conversation the memory was extracted from
  • Status: Whether the memory is currently enabled or disabled

Enabling / Disabling Memory

If a memory is no longer accurate or you don’t want Experts to use it, you can disable it. Once disabled, Experts will no longer reference that information, but the entry remains in your list and can be re-enabled at any time.
Re-enable a previously disabled memory, and Experts will reference it again in future conversations.

Deleting a Memory

For memory entries you no longer need at all, you can permanently delete them. Deletion is irreversible.
Deleting a memory is irreversible. If you only want to temporarily prevent Experts from using a memory, use “Disable” instead of “Delete.”

Personal Memory vs Expert Memory

Profy has two memory systems with different scopes: The two types of memory complement each other:
  • Personal memory tells all Experts that you prefer TypeScript
  • Expert memory tells your coding Expert about the specific project you’re working on, the technical approaches you’ve discussed, and conclusions from prior conversations
See Expert Memory for details.

How Memory Affects Conversations

At the start of each conversation, the Expert loads your personal memory as contextual reference. This means:
  • Expert responses will better match your style and preferences
  • You won’t need to re-explain your project background every time
  • Experts will proactively avoid approaches you dislike
Memory serves as a reference, not a directive. In certain scenarios, an Expert may make choices that differ from your general preferences based on the task at hand—for example, if you prefer TypeScript but the current task involves modifying a Python project, the Expert won’t rewrite it in TypeScript.

Privacy

  • Your personal memory is visible and manageable only by you
  • Other users and Expert creators cannot access your memory
  • Memory is not used for model training
  • You can disable or delete any memory entry at any time
  • Using Temporary Chat mode completely skips memory reading and writing