Honcho is an open-source memory and user-modeling backend for AI agents that reasons over conversations to maintain evolving user context across sessions, tools, and machines.
Why it matters
API Docs
API Available
overview
Honcho is an open-source AI memory tool that enables developers and businesses creating stateful AI agents to build persistent, evolving models of users from conversations. It supports memory across sessions, tools, and machines, using conversation reasoning to retain preferences and context. Honcho is provided by Plastic Labs and is designed as a memory and user-modeling backend rather than a full agent runtime.
features
Honcho provides conversation-derived memory and user modeling for stateful AI agents. Its documented technical scope includes text input, Plastic Labs Neuromancer models, an API, and MCP integration.
use cases
Honcho is intended for developers and businesses building AI agents that need user context to persist beyond a single conversation. The listed application areas include customer support, education, gaming, productivity, and AI companionship.
how to use
Developers can begin with the Honcho API documentation at https://honcho.dev/docs/ and connect an agent through the API or MCP. Configure the memory and reasoning usage to suit the application; the published pricing distinguishes memory ingestion from reasoning queries.
pricing
Honcho is listed as freemium, with usage prices for memory ingestion and reasoning. Memory is listed at $2.00 per million tokens ingested. Reasoning tiers are priced per query; the available facts do not specify the included free-tier allowance or whether these prices are current, so confirm rates on Honcho’s website before purchase.
Enjoying this? Get one like it in your inbox each morning.
one email a day · unsubscribe in two clicks · no third-party tracking
Similar Tools
Honcho is positioned as a plug-in memory and user-modeling backend. The supplied comparisons distinguish it from tools focused on fact extraction, agent runtime management, or knowledge-graph construction; implementation details and product capabilities may vary by provider.
Provides a multi-layered extraction pipeline that dynamically extracts, scores, and updates discrete user facts and preferences without manual graph curation.
Mem0 is faster to plug into existing agent stacks like LangChain or AutoGen, but Honcho gives you deeper dialectic user modeling and theory-of-mind representation beyond simple key-value fact extraction.
Manages agent memory like an operating system with tiered context windows, editable memory blocks, and git-backed context storage.
Letta is a full stateful agent runtime (formerly MemGPT) that dictates the agent's reasoning loop, whereas Honcho acts purely as a plug-and-play memory and user-modeling backend.
Constructs an evolving, temporally aware knowledge graph directly from conversational turns and external business data.
Graphiti provides clear entity-relation provenance over time, but you have to self-host Neo4j and write custom extraction logic rather than consuming Honcho's out-of-the-box user-profiling APIs.
Combines temporal knowledge graphs with hybrid vector and BM25 search to serve low-latency chat history and automated session summaries.
Zep excels at enterprise context enrichment and temporal fact invalidation, but its primary platform is closed-source managed SaaS, unlike Honcho's open-source architecture.
More on Stork
Other tools in this category, matched by shared tags
One short daily email of tools worth shipping. No drip funnel.
one email a day · unsubscribe in two clicks · no third-party tracking
For builders
AI agents read it. Buyers land on it. It answers in eight languages and over MCP. Your tool can have one like it — live in 24 hours.
