Memocore Knowledge Base

0 followers

Your guide to Memocore. Memocore is your persistent memory for AI — it saves notes, ideas, and knowledge so they stay with you across chats and tools. Here you'll learn what Memocore does and how it works, discover its features and best tips, and see real ways to use it in everyday work. Browse the articles or search for answers to the most common questions.

Shared project

This is Memocore

Memocore is the memory your AI clients share. Save something once and every assistant you use already knows it — nothing to re-explain, nothing to copy between chats.

Rolling out Memocore in your company, step by step

A practical rollout: create the company on Team, start with one project, seed it with the knowledge you re-explain most, invite the team by role, connect everyone's AI client, then split and publish.

アンアン株式会社
Jul 16, 2026
teamsuse-casesgetting-started

How to bring Memocore into a company in practice. The whole rollout takes an afternoon, and it works best when you start with one team and one painful topic rather than trying to move everything at once. Step 1 - create the company and pick the Team plan. The Team plan ($15 per seat per month) is the only one with companies, members and per-seat billing. The person who creates it becomes the owner: they manage members, billing and API keys. Step 2 - start with ONE project, not ten. Pick the knowledge your team re-explains most often - conventions, decisions, "how we do X". A project is a shared space whose memos every member can see. Give it a clear description: the description is what routes new memories to the right project later, so write it as "what belongs in here". Step 3 - seed it with real knowledge. The fastest start is to move what already exists: architecture decisions, coding conventions, policies, the onboarding checklist, the answers your team keeps repeating in chat. One coherent topic per memo. Attach the source documents as files if you have them. Step 4 - invite the team by email and set roles. Each project member gets a role: owner, editor (full read/write - the default) or viewer (read-only). Invites expire after 7 days; each accepted member takes one seat. Step 5 - have everyone connect their own AI client. Each person connects Claude, ChatGPT, Cursor, VS Code or Claude Code to the same memory over MCP. That is the moment it clicks: people stop re-briefing their assistant and stop asking colleagues questions the memory already answers. From then on they save knowledge just by talking to their assistant ("remember that we..."). Step 6 - expand by splitting projects. As it grows, split by team or domain rather than piling everything into one project. Company projects are isolated, so members only see the projects they are in - that is how you keep engineering, sales and legal separate. Step 7 - open it up where it makes sense. Publish a project behind a public link so customers, contractors or the community can read it (public readers take no seat and cost nothing), and mint a read-only API key to put your own assistant or support bot on top of that memory. Practical advice: appoint one owner per project who keeps it tidy; write memos as answers to questions people actually ask, not as documentation nobody reads; and let it grow from use instead of running a migration project. Memo history shows what changed, when and by whom, so nothing quietly rots. For enterprise needs (SSO, security review, invoicing), contact [email protected].