Release Date: July 2025

Feature Status: Beta

Default State: Disabled

Feature Summary

LLMs in HyperFlow can now store and retrieve persistent information across conversations without requiring manual tool node setup.

This enables chatbots and AI assistants to remember user preferences, past interactions, and important data between sessions.

What's New

Memory Tools

When enabled on a Call LLM or Call LLM with tools node, the following tools become automatically available to the LLM:

• store_memory: Save or update information with a key
• retrieve_memory: Get previously stored information
• list_memory_keys: See all available memory keys
• delete_memory: Remove stored information

Key Features

• Three persistence scopes: Project-wide (default), flow-specific ($flow.), session-specific ($session.)
• Automatic key discovery: LLMs see recently used keys in their context to help
• Simple updates: Storing with an existing key automatically overwrites
• Entity isolation (NEW): Optional entityID-based data isolation for multi-user/multi-tenant scenarios, giving true per-user long-term memory to LLM-based applications

Why Entity Isolation Matters

The entityID feature solves a critical challenge in modern AI applications: providing consistent per-user long-term memory across different LLM services. Traditional approaches require: