pip install mnemo-ai

Persistent Memory
for AI Agents

Inforge builds infrastructure that lets agents remember and share knowledge across sessions, across conversations, across teams. Our first product is Mnemo — a structured memory system purpose-built for AI agents.

Explore Mnemo View on GitHub

Copied! $ pip install mnemo-ai

Copied! $ pip install mnemo-ai[mcp]

Product

Mnemo — Persistent Memory for Agents

Agents accumulate knowledge — observations, decisions, learned context. Without a memory layer, that knowledge evaporates between sessions. Mnemo is a structured memory system purpose-built for agents: queryable, persistent, and designed for tool-calling loops.

Remember

Store typed knowledge — episodic, semantic, procedural — with Bayesian confidence scoring that evolves as evidence accumulates.

Recall

Semantic search with composite similarity-confidence ranking. Temporal anchoring, domain tags, and tunable verbosity controls.

Agent-to-agent memory sharing with explicit capability controls. The owner decides what gets shared and with whom — no blanket access.

82.1%

Second overall on the LoCoMo benchmark. First in multi-hop reasoning.
Validated against the leading long-conversation memory evaluation, competitive with other leading memory systems.

# MCP integration — works with Claude Desktop, any MCP-compatible client
# Install the server:
$ uvx mnemo-mcp

# Or use the Python client directly:
from mnemo import MnemoClient

client = MnemoClient("https://your-server.com")

# Store a memory
client.remember(
    content="User prefers async architectures over synchronous RPC",
    domain_tags=["architecture", "preferences"]
)

# Recall with semantic search
results = client.recall(
    query="What architectural patterns does the user prefer?",
    max_results=5
)

# Share memories with another agent
client.share(
    query="architecture preferences",
    share_with="planning-agent:team.acme"
)

Team

Built by Practitioners

Tom Davis

Co-founder

Quantitative research at FactSet and CFA charterholder with a PhD in theoretical physics. Twenty years in quantitative finance. Builds the systems.

Twitter Bluesky LinkedIn Soliton

Nels Ylitalo