general-intellect 0.1.0

Persistent, semantic memory for AI agents. No vector DB required.

general-intellect (Python SDK)

Persistent, semantic memory for AI agents. No vector DB required.

Install

pip install general-intellect

Usage

from generalintelect import GIClient

gi = GIClient(url="http://localhost:8000", agent_id="support-bot")

# In your LLM call handler:
context = gi.recall(query=user_message, namespace=call_id)
response = llm.complete(system_prompt + context + user_message)
gi.remember(content=f"User said: {user_message}", namespace=call_id)

Async

from generalintelect import AsyncGIClient

gi = AsyncGIClient(url="http://localhost:8000", agent_id="support-bot")

context = await gi.recall(query=user_message, namespace=call_id)
await gi.remember(content=user_message, namespace=call_id)

Vapi / Retell middleware

from generalintelect import GIClient
from generalintelect.vapi import with_memory

gi = GIClient(agent_id="support-bot")  # reads GI_URL from env

@with_memory(gi)
def handle_webhook(payload: dict) -> dict:
    context = payload["_gi_context"]  # injected automatically
    response = llm.complete(system_prompt + context + transcript)
    return {"response": response}

Memory degrades gracefully — if GI_URL is not set, recall() returns "" and remember() is a no-op.

Run the memory server

docker run -p 8000:8000 generalintelect/gi-server