ContextNest is governed context for your AI: your company's knowledge as structured, graph-linked Markdown, versioned and approved by the people who own it. Any MCP-compatible AI (Claude, Cursor, Antigravity) reads directly from it, so every tool works from the same approved version. The Desktop app is free and runs locally. Patent pending.

ContextNest handles living knowledge: structured, versioned, human-approved documents your AI reads directly, so the same question pulls the same context every time. RAG handles static knowledge: retrieval over large corpora that rarely change. PromptOwl governs both. Our research preprint (not peer reviewed), with co-authors from IBM Research and Emory University, measured a 97% answer pass rate versus 90–93% for keyword search (BM25) in a stale-version test, at about a third of the input tokens.

PromptOwl adds stewardship to ContextNest — a review workflow so context gets approved before your AI treats it as true, and role-based access so you control who can see it. On top of that: agents built by chatting, an audit trail of conversations, a record of every read, and every change to your knowledge, and every major model included in your subscription.

No. Your data is never used to train models. Prompts go only to the model providers you choose. With ContextNest Desktop, your files never leave your machine at all. Your knowledge stays in plain, portable files on infrastructure you choose — yours permanently.

The ContextNest CLI, engine and MCP server are open source (AGPL-3.0) and free. Install the CLI and it runs locally against your Markdown vault. The team workspace is free for up to 10 users, on our cloud or your servers. Hosted is $10 per user per month on our cloud. Self-Hosted is $10 per user with a 30-user minimum, unlimited storage and no overage bills. Every paid plan starts with a free month. Enterprise is custom, on one annual contract. Your own AI tools reading a nest over MCP never use credits. Credits only apply when PromptOwl supplies the model.

The ContextNest CLI, engine and MCP server are open source (AGPL-3.0) and stay free on every plan. Install the CLI locally and it runs against your Markdown vault, invisible in the background, with any MCP-compatible AI. PromptOwl is what you add when memory needs to work across a team instead of one person: agents built by chatting, role-based access, approval workflows, a hash-chained history of every change, and every major model included in one subscription. ContextNest is what your AI knows. PromptOwl is what it can do — and who controls it.

Every plan runs on the same SOC 2 and HIPAA controls. SOC 2 certification and reporting and a HIPAA BAA come with Enterprise. Role-based access, approval before AI may use a document, and a record of every read are part of the product.

Search finds things. This governs what your AI may treat as true. Enterprise search tools index everything your organization has and let AI search it: comprehensive, but ungoverned. ContextNest governs which knowledge is authoritative, approved and current, so when your audit team asks what was approved, and who authorized it, there's an answer.