Skip to main content

​
Setup

1

Create and activate a virtual environment

uv venv --python 3.12
source .venv/bin/activate
2

Install dependencies

uv pip install -U agno openai sqlalchemy chromadb
3

Export your API key

export OPENAI_API_KEY=sk-***

​
Create an Agent

basic_agent.py
from agno.agent import Agent
from agno.models.openai import OpenAIResponses

agent = Agent(model=OpenAIResponses(id="gpt-5.2"))

agent.print_response("Hi! I'm Alice. I work at Anthropic as a research scientist.", stream=True)

python basic_agent.py
That’s a basic agent. It responds, but it doesn’t remember anything.

​
Add Learning

Add a database and set learning=True:
01_always_learn.py
from agno.agent import Agent
from agno.db.sqlite import SqliteDb
from agno.models.openai import OpenAIResponses

agent = Agent(
    model=OpenAIResponses(id="gpt-5.2"),
    db=SqliteDb(db_file="tmp/agents.db"),
    add_history_to_context=True,
    learning=True,
    markdown=True,
)

if __name__ == "__main__":
    user_id = "[email protected]"

    # Session 1: Share information naturally
    print("\n--- Session 1: Extraction happens automatically ---\n")
    agent.print_response(
        "Hi! I'm Alice. I work at Anthropic as a research scientist. "
        "I prefer concise responses without too much explanation.",
        user_id=user_id,
        session_id="session_1",
        stream=True,
    )
    lm = agent.get_learning_machine()
    lm.user_profile_store.print(user_id=user_id)
    lm.user_memory_store.print(user_id=user_id)

    # Session 2: New session - agent remembers
    print("\n--- Session 2: Agent remembers across sessions ---\n")
    agent.print_response(
        "What do you know about me?",
        user_id=user_id,
        session_id="session_2",
        stream=True,
    )

python 01_always_learn.py
Session 2 is a new conversation. Different session ID. But the agent remembers Alice. How it works: When you set learning=True, the agent extracts user profile and memories in parallel with generating the response, stores them in the database, and recalls them in future sessions. No tool calls visible. The agent learns in the background.

​
Agentic Learning

For more control, use agentic mode. The agent receives tools and decides what to save.
02_agentic_learn.py
from agno.agent import Agent
from agno.db.sqlite import SqliteDb
from agno.learn import (
    LearningMachine,
    LearningMode,
    UserMemoryConfig,
    UserProfileConfig,
)
from agno.models.openai import OpenAIResponses

agent = Agent(
    model=OpenAIResponses(id="gpt-5.2"),
    db=SqliteDb(db_file="tmp/agents.db"),
    add_history_to_context=True,
    learning=LearningMachine(
        user_profile=UserProfileConfig(mode=LearningMode.AGENTIC),
        user_memory=UserMemoryConfig(mode=LearningMode.AGENTIC),
    ),
    markdown=True,
)

if __name__ == "__main__":
    user_id = "[email protected]"

    # Session 1: Agent decides what to save via tool calls
    print("\n--- Session 1: Agent uses tools to save profile and memories ---\n")
    agent.print_response(
        "Hi! I'm Alice. I work at Anthropic as a research scientist. "
        "I prefer concise responses without too much explanation.",
        user_id=user_id,
        session_id="session_1",
        stream=True,
    )
    lm = agent.get_learning_machine()
    lm.user_profile_store.print(user_id=user_id)
    lm.user_memory_store.print(user_id=user_id)

    # Session 2: New session - agent remembers
    print("\n--- Session 2: Agent remembers across sessions ---\n")
    agent.print_response(
        "What do you know about me?",
        user_id=user_id,
        session_id="session_2",
        stream=True,
    )

python 02_agentic_learn.py
Watch for tool calls in the output. The agent explicitly decides when to save.

​
Cross-User Insights

Agents can learn knowledge that transfers across users. One user teaches the agent, another benefits.
03_learned_knowledge.py
from agno.agent import Agent
from agno.db.sqlite import SqliteDb
from agno.knowledge import Knowledge
from agno.knowledge.embedder.openai import OpenAIEmbedder
from agno.learn import LearnedKnowledgeConfig, LearningMachine, LearningMode
from agno.models.openai import OpenAIResponses
from agno.vectordb.chroma import ChromaDb, SearchType

knowledge = Knowledge(
    name="Agent Learnings",
    vector_db=ChromaDb(
        name="learnings",
        path="tmp/chromadb",
        persistent_client=True,
        search_type=SearchType.hybrid,
        embedder=OpenAIEmbedder(id="text-embedding-3-small"),
    ),
)

agent = Agent(
    model=OpenAIResponses(id="gpt-5.2"),
    db=SqliteDb(db_file="tmp/agents.db"),
    add_history_to_context=True,
    learning=LearningMachine(
        knowledge=knowledge,
        learned_knowledge=LearnedKnowledgeConfig(mode=LearningMode.AGENTIC),
    ),
    markdown=True,
)

if __name__ == "__main__":
    # Session 1: User 1 teaches the agent
    print("\n--- Session 1: User 1 saves a learning ---\n")
    agent.print_response(
        "We're trying to reduce our cloud egress costs. Remember this.",
        user_id="[email protected]",
        session_id="session_1",
        stream=True,
    )
    lm = agent.get_learning_machine()
    lm.learned_knowledge_store.print(query="cloud")

    # Session 2: User 2 benefits from the learning
    print("\n--- Session 2: User 2 asks a related question ---\n")
    agent.print_response(
        "I'm picking a cloud provider for a data pipeline. Key considerations?",
        user_id="[email protected]",
        session_id="session_2",
        stream=True,
    )

python 03_learned_knowledge.py
User 2 gets advice informed by User 1’s insight. No shared context. No explicit handoff.

​
Next

Your First Multi-Agent System

Build your first multi-agent system.

Build Agents

Configuration, tools, and advanced patterns