Pics — visual notes, drawn by AI

Every diagram here is a real NoteBrain AI note. None of them were drawn by hand — an AI agent generated each one inside NoteBrain AI over MCP, from a one-line prompt like “make a one-slide summary of OAuth2 based on my powerpoint template in NoteBrain AI”.

MCP + NoteBrain AI Plug your AI agent into your second brain — read, search & write notes ① You ask — in plain words > Make a 101 course on photosynthesis, with an SVG diagram …a note · a course · a slide ② Agent connects 🤖 AI Agent Claude Code · Codex · Gemini CLI asks MCP · HTTP 🔒 call → result 🔌 NoteBrain.ai · MCP Server (*) speaks the Model Context Protocol ③ Tools the agent can call 🔎 search_notes 📖 read_note ✍️ create_note 🗂 organize 📋 create_kanban 💬 ask_notes read · search · write — across all 7 editor types (*) 💻 Free on Desktop MCP server (token-secured) OR ☁️ Cloud NoteBrain AI MCP server writes ④ Your second brain 📂 Photosynthesis 101 📄 Introduction 🧩 How It Works 📄 Light & Dark Reactions 📄 Key Terms created live by your agent ✅ Notes appear instantly Ask in your terminal — watch it appear in your brain Built with NoteBrain.ai
MCP + NoteBrain AI
RAG Retrieval-Augmented Generation — answer using YOUR data ① Index your docs (once, offline) 📄Docs ✂️Chunks 🧠Embedding model split into passages text → vector (numbers) [0.12, -0.74, 0.33 …] store ⬇ Vector DB 🛢 embeddings stores millions of vectors 🔎 find top-k nearest ② At question time Question 🧠Embed vector space near = similar meaning (cosine) ③ Generate the answer 📚Top-k chunks (context) + your question retrieved 🤖 LLM Gemini · GPT · Claude ✅ Grounded answer grounded in YOUR data → fewer hallucinations · cites sources A picture is worth a thousand words Built with NoteBrain.ai
RAG — Retrieval-Augmented Generation
OAuth2 Let an app use your data — without ever sharing your password 🌐 App the client 🔐 Auth Server Google · GitHub · Okta 📦 Resource API your protected data ① redirect: "Login with…" + consent ② ↩ authorization code (one-time) ③ code + client_secret 🔒 (back-channel) ④ access token 🔑 (+ refresh token) ⑤ call API — Authorization: Bearer <token> ⑥ ✅ protected data returned ✅ The app acts on your behalf — your password is never shared with it A picture is worth a thousand words Built with NoteBrain.ai
OAuth2 — Authorization Code Flow
ChatBot vs Agent a chatbot replies once · an agent loops: think → act → observe → repeat ① ChatBot — one prompt, one answer 🙋 You · a question 🧠 LLM generate text 💬 Reply one shot · no memory · no tools ② Agent — a loop with memory & tools 🎯 Goal 🧠 Model reason · plan reason ⟳ act ⟳ observe act observe done ✅ Goal done 💾 Memory 🧩 Skills 🔌 MCP 🛠 Tools calls these, reads the results, then loops — until the goal's met A picture is worth a thousand words Built with NoteBrain.ai
ChatBot vs Agent

Want your own? Connect your AI agent to NoteBrain AI over MCP and ask it to draw one. Start free →