Mnemo

What Mnemo does

Mnemo is a US-hosted, founder-led company that sells long-term memory infrastructure for AI agents. Its core product is a REST API that ingests conversations, documents, and external data sources, extracts atomic facts, entities, and timestamps, and returns cited, prompt-ready context in response to natural-language queries. Retrieval is powered by a proprietary seven-signal hybrid architecture (vector similarity, BM25 keyword search, atomic fact, temporal reasoning, entity graph, concept boosting, and semantic bridge expansion) fused with intent-aware Reciprocal Rank Fusion, which Mnemo reports at 85.2% accuracy and 99.2% Recall@5 on LongMemEval-S with ~150ms warm search latency.

Mnemo monetizes through a four-tier subscription model (Free, Pro at $19/month, Team at $99/month, and custom Enterprise), supplemented by opt-in usage-based overage for ingestion. Go-to-market is API-first and product-led: developers onboard through a self-serve dashboard, install SDKs from npm and PyPI, or consume the service through a hosted MCP server compatible with Claude Desktop, Cursor, Windsurf, VS Code, and Zed. Native integrations span eight agent frameworks (LangChain, LangGraph, Vercel AI SDK, AutoGen, CrewAI, PydanticAI, LlamaIndex, Mastra) and 20 data-source connectors (Google Drive, Gmail, Notion, Slack, HubSpot, GitHub, Linear). Packaged product extensions include Mnemo Company Brain for Slack and the Ask Mnemo embeddable widget.

The company is privately held by founder Shah Hussain, with no disclosed institutional funding and only one named team member. Revenue, headcount, and customer counts are not disclosed; SOC 2 Type 2, HIPAA, ISO 27001, EU data residency, and self-hosting via Docker Compose or Helm are stated as roadmap items.

Name

Mnemo

Legal name

Mnemo

Website

https://mnemohq.com

Company type

Private

Operating status

Operating

Short description

Mnemo provides long-term memory infrastructure for AI agents via a REST API and SDKs that store conversations and documents as atomic facts and return cited, prompt-ready context through a seven-signal hybrid retrieval engine.

Ownership category

akta.pro rank

Product category

AI Agent Memory Infrastructure

NAICS

Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518)

SIC

Services-Computer Programming, Data Processing, Etc. (7370)

akta.pro primary industry

LLMOps & Generative AI Platforms (Prompt/Agent Orchestration, RAG) (HDAEANAD)

Keywords

• AI agent memory
• Long-term memory infrastructure
• Hybrid retrieval API
• RAG memory platform
• MCP server tools

GTM type

B2B

Offering type

Software

Cost components

Technology or R&D, Infrastructure, Personnel, Marketing or Sales, Operations, Supply Chain

Revenue model

1. Subscription Tiers: Four-tier subscription model: Free (no cost), Pro ($19/mo), Team ($99/mo), and Enterprise (custom pricing). Paid tiers include usage allocations for ingestion units, standard retrievals, and YouTube video processing.
2. Usage-Based Overage: Ingestion overage at $4 per 1,000 units — only charged after enabling and setting a finite ceiling. Standard retrievals and advanced retrieval (LLM reranking, precise mode) have separate usage计量.
3. Freemium Entry: Free tier provides 1,000 ingestion units, 10,000 standard retrievals, 30 min YouTube transcript, 10 min YouTube visuals per month. No credit card required to start.

Pricing tiers

Go-to-market motion2 records

Distribution channels3 records

Marketing channels5 records

Core offering

Mnemo provides long-term memory infrastructure for AI agents via a REST API and SDKs that store, retrieve, and serve cited memory facts. Its hybrid retrieval fuses seven signals (vector, BM25, atomic fact, temporal, entity graph, concept boost, semantic bridge) using intent-aware Reciprocal Rank Fusion, exposed through a hosted MCP server, framework integrations (LangChain, LangGraph, Vercel AI SDK, AutoGen, CrewAI, etc.), and 20 data-source connectors. Core product is sold on a freemium subscription model with usage-based overages and a custom-priced enterprise tier.

Product overview

Mnemo offers a unified long-term memory infrastructure platform for AI agents, consisting of the core Mnemo Memory API with hybrid retrieval (seven fused signals), supported by multiple SDK packages in TypeScript (npm), Python (PyPI), and CLI. The platform includes specialized packages for major agent frameworks (LangGraph, Vercel AI SDK, Anthropic, Mastra), an MCP server for Claude/Cursor/VS Code/Zed, and the Company Brain Slack app for team memory. Additional modules include the Ask Mnemo embeddable widget and 20 data-source connectors for external tool sync.

Differentiator

Problem solved

Functional benefit

Products and services

• Mnemo Memory API Long-term memory infrastructure for AI agents providing hybrid retrieval across seven signals (vector, BM25, atomic fact, temporal, entity graph, concept boost, semantic bridge) fused with intent-aware Reciprocal Rank Fusion; achieves ~150ms warm search p50 on the production API and is sold as a freemium API product with subscription tiers.
• Mnemo TypeScript SDK (getmnemo) Zero-runtime-dependency TypeScript SDK that exposes the full Mnemo Memory API as typed functions, installable via npm, targeting developers integrating memory into AI agent applications.
• Mnemo Python SDK (getmnemo) Python SDK on PyPI for integrating Mnemo long-term memory storage and retrieval into Python-based AI agents, frameworks, and backends.
• Mnemo CLI (getmnemo-cli) Command-line interface that lets developers add, search, and inspect Mnemo memories from a terminal or CI pipeline; installable globally via npm.
• Mnemo MCP Server (getmnemo-mcp) Hosted and local Model Context Protocol server that exposes Mnemo search, add, update, and delete as MCP tools for Claude Desktop, Cursor, Windsurf, VS Code, and Zed.
• Mnemo Company Brain Slack app that learns from public channel messages and answers team questions with cited sources linking back to the original Slack messages, sold as an add-on to Mnemo Memory.
• Ask Mnemo Widget Embeddable website widget that lets visitors ask questions against a site's docs, help center, and changelog, returning numbered answers with source citations; installed via a single script tag.
• Data-Source Connectors Suite of 20 managed connectors to external tools such as Google Drive, Gmail, Notion, Slack, HubSpot, GitHub, and Linear that continuously backfill, update incrementally, and propagate deletions into the same hybrid search as API-written memories.
• Mnemo Subscription Plans (Free, Pro $19/mo, Team $99/mo, Enterprise) Commercial subscription tiers for accessing Mnemo Memory, with usage allocations for ingestion units, standard retrievals, and YouTube processing; Enterprise tier adds SSO/SAML, custom audit retention, SLAs, and optional self-hosting.

Quantifiable outcome

• 85.2% accuracy on LongMemEval-S benchmark (vs. lower for pure-vector approaches)
• +3 more outcomes

Segments2 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration18 records

AI capability4 records

Feature8 records

Partnerships

Five partnerships are on record, tiered core.

• LangChaincoreTechnology or IntegrationFirst-class LangChain integration via getmnemo package. Supports decorated tool functions in Python for seamless memory storage and retrieval within LangChain agent workflows.
• LangGraphcoreTechnology or IntegrationGraph nodes for recall and store in LangGraph state graphs via getmnemo-langgraph package. Enables persistent memory across graph execution cycles.
• Vercel AI SDKcoreTechnology or IntegrationTyped Zod tool definitions via getmnemo-vercel-ai for streamText, generateText, and useChat in the Vercel AI ecosystem.
• Anthropic (MCP)coreTechnology or IntegrationMnemo MCP server exposes memory tools for Claude Desktop. OAuth 2.0 authorization server for MCP clients with PKCE. Remote MCP endpoint at mcp.mnemohq.com/mcp.
• Data Source Connectors (20 sources)coreTechnology or Integration20 data-source connectors including Google Drive, Gmail, Notion, Slack, HubSpot, GitHub, Linear with continuous backfill, incremental updates, and deletion propagation.

Scale indicators8 records

Recent moves6 records

Expansion highlights6 records

Direct peers

• Mem0: Mem0 provides a dedicated long-term memory layer for LLM-based agents with self-improving memory and framework integrations. It is the most direct head-to-head competitor to Mnemo's positioning as memory infrastructure for AI agents.
• Zep (Zep AI): Zep offers a long-term memory and context engineering platform for AI assistants and agents, with similar hybrid retrieval and framework integrations. Competes directly with Mnemo on memory APIs for production agent workloads.
• Letta: Letta (formerly MemGPT) builds persistent memory and state management for LLM agents and chatbots. It overlaps Mnemo's core value proposition of giving agents durable, queryable memory across sessions.
• Cognee: Cognee is an open-source memory layer for AI applications combining vector search and knowledge graphs for agent context. Comparable to Mnemo's entity-graph-augmented hybrid retrieval approach.

Broad incumbents

• Pinecone: Pinecone is a managed vector database widely used as the retrieval backbone for AI applications. It is a broad incumbent whose vector infrastructure overlaps with Mnemo's retrieval stack, though Pinecone lacks Mnemo's atomic-fact and intent-routing layer.
• Weaviate: Weaviate is an open-source vector and hybrid-search database with built-in BM25 and vector fusion. It is a broad incumbent offering retrieval primitives that overlap with Mnemo's hybrid retrieval, but at the database layer rather than as an agent memory product.
• Qdrant: Qdrant is a high-performance vector database with hybrid search and metadata filtering. It competes as underlying retrieval infrastructure for the same agent-memory workloads Mnemo targets, but does not provide memory orchestration or atomic-fact extraction.
• Chroma: Chroma is an open-source embedding database commonly used for LLM applications and agent memory. It overlaps Mnemo's retrieval role but operates at the embedding-store layer without Mnemo's seven-signal fusion or agent-framework integrations.

Emerging players

• LangChain: LangChain offers its own memory modules and LangGraph-based state stores that compete with Mnemo's framework integrations. As the dominant agent framework, LangChain represents both a partner (Mnemo integrates with it) and a competitive threat if it deepens native memory features.
• LlamaIndex: LlamaIndex provides RAG and memory components for LLM applications and integrates natively with agent frameworks. It overlaps Mnemo's retrieval and indexing role, particularly for knowledge-base and user-memory hybrid use cases.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks6 records

Key highlights7 records

Customer concentration

• GitHubgithub.com/shhahhussain

Compliance5 records

Revenue estimate

Valuation estimate

Number of profiles

Profiles1 record

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Mnemo