Lians

Bitemporal long-term memory for AI agents. Keep current facts clean, reconstruct what an agent knew at a past time, and retain tamper-evident audit records.

Install

pip install lians-sdk
pip install lians-sdk[local]         # SQLite plus real local semantic embeddings
pip install lians-sdk[mcp]           # Local MCP server
pip install lians-sdk[langchain]     # LangChain
pip install lians-sdk[langgraph]     # LangGraph
pip install lians-sdk[crewai]        # CrewAI
pip install lians-sdk[openai-agents] # OpenAI Agents SDK
pip install lians-sdk[autogen]       # AutoGen v0.4
pip install lians-sdk[all]           # Everything

Quickstart

from datetime import datetime, timezone
from lians import LocalLiansClient

mem = LocalLiansClient()  # No server, Docker, or API key

mem.add(
    agent_id="analyst-1",
    content="NVDA FY2026 revenue guidance raised to $40B",
    event_time=datetime(2025, 11, 19, 16, tzinfo=timezone.utc),
    metadata={"ticker": "NVDA", "metric": "revenue_guidance"},
    importance=0.9,
)

# Superseded facts are excluded before they reach the model
current = mem.recall(agent_id="analyst-1", query="NVDA revenue guidance")

# Reconstruct what was known on a past date
past = mem.recall_at(
    agent_id="analyst-1",
    query="NVDA revenue guidance",
    as_of=datetime(2025, 3, 1, tzinfo=timezone.utc),
)

Why Lians

• Bitemporal facts with event time and ingestion time
• Deterministic supersession before memories reach the model
• Point-in-time recall and lookahead-bias checks
• Tamper-evident audit history and a crypto-erasure workflow
• Local SQLite mode with no server or API key
• Hosted and self-hosted deployment paths

See the published benchmark results, regulated-memory evaluation, and public correction ledger. The evaluation includes runnable adapters so results can be reproduced and challenged.

Framework integrations

from lians.langchain_integration import LiansChatHistory, build_tools
from lians.langgraph_integration import create_recall_node, create_remember_node
from lians.crewai_integration import build_crewai_tools
from lians.openai_agents_integration import build_openai_agent_tools
from lians.autogen_integration import build_autogen_tools

Hosted or self-hosted API

from lians import LiansClient

mem = LiansClient(base_url="https://mem.yourfirm.internal", api_key="...")

Full documentation: github.com/Lians-ai/Lians