Learn Config

Configure continuous learning to capture patterns and preferences from agent interactions using LearnConfig

python

from praisonaiagents import Agent, LearnConfig

agent = Agent(
    name="Personal Assistant",
    instructions="You are a helpful assistant that remembers user preferences.",
    learn=LearnConfig(
        persona=True,
        insights=True,
        mode="agentic",
    )
)

agent.start("I prefer concise answers with bullet points")

Agents learn from every interaction — capturing user preferences, insights, and patterns to improve future responses automatically.

The user interacts with the agent; learnings are extracted and stored, then recalled in future sessions.

Diagram

Quick Start

Enable Learning with Defaults

python

from praisonaiagents import Agent

agent = Agent(
    name="Learning Agent",
    instructions="You are a helpful assistant",
    learn=True
)

agent.start("I work in Python and prefer type hints")

Configure Learning Capabilities

python

from praisonaiagents import Agent, LearnConfig

agent = Agent(
    name="Smart Assistant",
    instructions="Remember my preferences and insights.",
    learn=LearnConfig(
        persona=True,
        insights=True,
        patterns=True,
        mode="agentic",
    )
)

agent.start("Always format code examples with type annotations")

Use a Database Backend

python

from praisonaiagents import Agent, LearnConfig

agent = Agent(
    name="Persistent Learner",
    instructions="Store learnings in a database.",
    learn=LearnConfig(
        backend="sqlite",
        db_url="sqlite:///learn.db",
        persona=True,
        insights=True,
    )
)

agent.start("My timezone is UTC+5:30")

How It Works

Diagram

Choose a Learning Mode

Diagram

Choose a Storage Backend

Diagram

Configuration Options

LearnConfig Python Reference

Full parameter reference for LearnConfig

Option	Type	Default	Description
`persona`	`bool`	`True`	Capture user preferences and profile
`insights`	`bool`	`True`	Capture observations and learnings
`thread`	`bool`	`True`	Capture session/conversation context
`patterns`	`bool`	`False`	Capture reusable knowledge patterns
`decisions`	`bool`	`False`	Log decision history
`feedback`	`bool`	`False`	Capture outcome signals
`improvements`	`bool`	`False`	Self-improvement proposals
`mode`	`str`	`"disabled"`	`"disabled"`, `"agentic"`, or `"propose"`
`backend`	`str`	`"file"`	`"file"`, `"sqlite"`, `"redis"`, or `"mongodb"`
`db_url`	`str \	None`	`None`	Database connection URL (non-file backends)
`store_path`	`str \	None`	`None`	Custom path for file backend
`max_entries`	`int`	`0`	Max entries per store (0 = unbounded)
`retention_days`	`int`	`0`	Archive entries unused for N days (0 = keep forever)
`llm`	`str \	None`	`None`	LLM for learning extraction (defaults to agent’s LLM)
`scope`	`str`	`"private"`	`"private"` or `"shared"`
`nudge_interval`	`int`	`0`	Nudge agent every N turns. `0` = disabled; `Agent(learn=True)` sets it to `10`.
`nudge_min_tool_iters`	`int`	`3`	Tunable override, not a hard floor. `<= 0` fires the cadence on conversational (tool-less) turns too; a positive value gates on tool-work count.
`propose_skills`	`bool`	`False`	Enable `skill_manage` tool in propose mode

Common Patterns

Self-improving assistant with retention governance:

python

from praisonaiagents import Agent, LearnConfig

agent = Agent(
    name="Adaptive Assistant",
    instructions="Adapt to user preferences over time.",
    learn=LearnConfig(
        persona=True,
        insights=True,
        patterns=True,
        mode="agentic",
        max_entries=500,
        retention_days=90,
    )
)

Shared learnings across a team of agents:

python

from praisonaiagents import Agent, LearnConfig

shared_learn = LearnConfig(
    scope="shared",
    backend="sqlite",
    db_url="sqlite:///team_learnings.db",
    persona=True,
    insights=True,
)

researcher = Agent(name="Researcher", instructions="Research topics.", learn=shared_learn)
writer = Agent(name="Writer", instructions="Write content.", learn=shared_learn)

Best Practices

Start with learn=True before tuning

Enable learning with defaults first. Once you understand what gets captured, switch to LearnConfig to enable only the capabilities you need (persona, insights, patterns).

Use agentic mode for hands-off learning

Set mode="agentic" to have the agent automatically extract and store learnings after each conversation. This requires no manual action from users or developers.

Set retention limits in production

Use max_entries and retention_days to prevent unbounded growth. A limit of max_entries=1000 with retention_days=180 keeps the learning store manageable.

Use sqlite or redis for persistence across restarts

The default file backend stores JSON in ~/.praison. For production or multi-instance setups, use backend="sqlite" or backend="redis" with a db_url.

Memory Config

Session and persistent memory configuration

Agent Learning

Agent learn feature overview

Learning Retention

Govern how long learnings are kept

Learn Skill

Teach agents new skills dynamically

Bot Session Learning

Default-on cross-session learning for gateway/bot agents

Learn Config

Quick Start

How It Works

Choose a Learning Mode

Choose a Storage Backend

Configuration Options

Common Patterns

Best Practices

Related