Experia AI

The open-source experience learning layer for AI agents.

Experia enables agents to learn from past actions, failures, and outcomes. It provides experience capture, lesson extraction, behavioral improvement, and long-term cognitive memory without replacing existing agent frameworks.

Vision

Current AI agents start from zero on every interaction. Experia adds an experience learning loop around agents. It allows agents to remember what happened, understand why it worked or failed, and improve future decisions.

Observation → Action → Result → Experience → Lesson → Memory → Better Future Action

flowchart TD
    subgraph Multi-Agent Swarm
        AgentA[Coder Agent]
        AgentB[Researcher Agent]
        Supervisor[Supervisor Agent]
    end

    subgraph Experia AI Cognitive Layer
        Store[(Shared MemoryStore)]
        Eval[LLM Evaluator\nRoot Cause Analysis]
        RuleGen[Rule Generator]
        Reflect[Reflection Engine\nBatch Analysis]
        Ctx[Context Builder]
        
        AgentA & AgentB -- 1. Record Action --> Store
        Store -- 2. Evaluate Outcome --> Eval
        Eval -- 3. Extract Lesson (agent_role) --> Store
        Eval -- 4. Consolidate --> RuleGen
        RuleGen -- 5. Generate RULE --> Store
        Supervisor -- 6. Trigger reflect() --> Reflect
        Reflect -- 7. Generate Global STRATEGY --> Store
        Store -- 8. retrieve_context(agent_role) --> Ctx
    end
    
    Ctx -- 9. Inject Shared Knowledge --> AgentA & AgentB & Supervisor

Integrations

Experia acts as a cognitive plugin. It does not replace your agent frameworks (like LangChain, AutoGen, CrewAI), it enhances them by managing long-term memory, learned experiences, user knowledge, and behavioral patterns.

Getting Started

Experia uses an asynchronous and pluggable architecture. To use the advanced cognitive features (Root Cause Analysis, Rules, and Reflection), install with the llm extra. You can also install specific integration extras like langchain, langgraph, or openai:

pip install "experia[llm,langgraph]"

Note: You will need an OPENAI_API_KEY (or other litellm supported keys) exported in your environment.

For full documentation on classes, methods, and models, please see the API Reference.

Native LangGraph Integration

Experia provides native, stateful Nodes for LangGraph, the modern standard for Multi-Agent and Cyclical AI workflows.

import asyncio
from experia.core.learner import Learner
from experia.memory.store import SQLiteStore
from experia.experience.llm_evaluator import LLMEvaluator
from experia.integrations.langgraph.nodes import ExperiaContextNode, ExperiaLearningNode
from langgraph.graph import StateGraph, MessagesState


async def main():
    store = SQLiteStore("my_agent.db")
    await store.initialize()
    agent = Learner(store=store, evaluator=LLMEvaluator(model="gpt-4o-mini"))

    # Define your standard LangGraph
    builder = StateGraph(MessagesState)

    # 1. Add Experia Context Node (Injects learned knowledge before the agent acts)
    builder.add_node("inject_context", ExperiaContextNode(agent=agent))

    # 2. Add your Agent and Tool nodes
    # builder.add_node("agent", ...)
    # builder.add_node("tools", ...)

    # 3. Add Experia Learning Node (Extracts experiences after tools run)
    builder.add_node("learn", ExperiaLearningNode(agent=agent))

    # Flow
    builder.set_entry_point("inject_context")
    # builder.add_edge("inject_context", "agent")
    # builder.add_edge("agent", "tools")
    # builder.add_edge("tools", "learn")
    # builder.add_edge("learn", "agent")

    graph = builder.compile()

    # Now run your graph! Experia will automatically learn from every cycle.
    # await graph.ainvoke({"messages": [...]})


if __name__ == "__main__":
    asyncio.run(main())

Manual Core API

You can also use the core API manually without any frameworks:

import asyncio
from experia.core.learner import Learner
from experia.memory.store import SQLiteStore
from experia.experience.llm_evaluator import LLMEvaluator
from experia.improvement.rules import RuleGenerator


async def main():
    store = SQLiteStore("my_agent.db")
    await store.initialize()

    agent = Learner(
        store=store,
        evaluator=LLMEvaluator(model="gpt-4o-mini"),
        rule_generator=RuleGenerator(store=store, model="gpt-4o-mini"),
    )

    # Record actions explicitly
    await agent.record(
        task="Deploy web app",
        action="Restart Nginx",
        result="failed with config syntax error",
    )

    # Developer-controlled Nightly Reflection
    await agent.reflect(model="gpt-4o-mini", batch_size=50)


if __name__ == "__main__":
    asyncio.run(main())

License

This project is licensed under the MIT License. See the LICENSE file for details.