Skip to main content
Concepts Memory

​
Overview

Memory enables agents to remember conversations and learn about users. It maintains conversation history, creates summaries, and builds user profiles to provide context across interactions.
Storage Installation
pip install upsonic[storage]
This installs Upsonic with storage dependencies including SQLite, PostgreSQL, Redis, MongoDB, and Mem0 for memory management. These backends enable persistent conversation history, user profiles, and automatic memory management across agent sessions.

​
Key Features

  • Conversation History: Remember previous messages and maintain context
  • User Profiles: Learn user preferences and communication style
  • Automatic Summaries: Generate conversation summaries
  • Multiple Storage Backends: SQLite, Redis, PostgreSQL, or in-memory
  • Flexible Memory Types: Choose full history, summaries, or user analysis
  • Automatic Management: Memory is saved and loaded automatically

​
Example

from upsonic import Agent, Task, Memory from upsonic.storage.providers.sqlite import SqliteStorage  # Create storage and memory storage = SqliteStorage("sessions", "profiles", "memory.db") memory = Memory(  storage=storage,  session_id="session_001",  user_id="user_123",  full_session_memory=True,  summary_memory=True,  user_analysis_memory=True,  model="openai/gpt-4o" # Required for summary_memory and user_analysis_memory )  # Create agent with memory agent = Agent("openai/gpt-4o", memory=memory)  # First conversation task1 = Task("My name is Alice and I love Python") result1 = agent.do(task1)  # Second conversation - agent remembers task2 = Task("What's my name and favorite language?") result2 = agent.do(task2) print(result2) # Output: Your name is Alice and you love Python