Persistence Overview

Understanding agent state persistence

Persist agent state to resume sessions and maintain context across runs.

from praisonaiagents import Agent, Session

session = Session(session_id="my-session", persistence="sqlite")
agent = Agent(name="Assistant", session=session)

agent.start("Pick up where we left off.")

The user runs the agent again; stored memory and checkpoints avoid repeating earlier steps.

What Gets Persisted

• Memory: Conversation history and context
• Session State: Current task progress
• Checkpoints: Intermediate results
• Configuration: Agent settings

Quick Start

1. Simple Usage

   from praisonaiagents import Agent, Session
   
   # Create session with persistence
   session = Session(
       session_id="my-session",
       persistence="sqlite"
   )
   
   agent = Agent(
       name="Assistant",
       session=session
   )
   
   # Run agent
   agent.start("Hello!")
   
   # Later, resume the same session
   session = Session(session_id="my-session", persistence="sqlite")
   agent = Agent(name="Assistant", session=session)
   agent.start("What did we talk about?")  # Remembers previous context

2. With Configuration

   from praisonaiagents import Agent, Session
   
   session = Session(session_id="prod-session", persistence="postgresql")
   agent = Agent(name="Assistant", session=session, llm="gpt-4o-mini")
   agent.start("Continue from where we left off.")

Persistence Backends

CLI Session Resume

praisonai session resume <id> is a first-class restore — it brings back chat history, model, and agent name. Use a follow-up prompt to continue immediately:

praisonai session resume my-session "What did we discuss?"

See Session Command for the full reference including --transcript and cross-store lookup behaviour.

Related