miMind AI — Guide & Command Reference

Generate a Map

From the map list AI button

Tap the round pulsing AI button in your map list to create a brand-new map. Describe what you want, optionally pick a layout, theme and level of detail, attach files, images or links — then hit Generate and watch the map build itself node by node.

By default the map is built from concise node titles. Tick Generate Node Notes if you also want the AI to write longer explanatory notes on the nodes.

From a topic or description

A study plan for learning Spanish in 6 months, with weekly goals Business plan for a small coffee roastery — products, marketing, finances, suppliers

From pasted text

Paste meeting notes, an article, or any longer text — the AI organizes it into a structured map.

Turn these meeting notes into an action map: [paste your notes]

From a URL

Paste a link — the AI reads the page and maps its content.

https://en.wikipedia.org/wiki/Solar_System — map the main facts

With attachments

Use the + Add Files / Images button to attach material for the AI to build the map from. What you can attach:

PDFs — read page by page, including layout and tables. Up to 2 per request, 8 MB each.
Images (PNG, JPEG, GIF, WebP) — analyzed with vision. Up to 4 per request.
Text and code files — .txt, .md, .csv, .json, source files and similar; the first ~50 KB of each is read.
Links — the page is fetched and mapped alongside everything else.

Attach several files at once and the AI gives each distinct component its own root node, so you get side-by-side maps in one document rather than one crowded tree.

Video, archives, Word and Excel files can’t be read — those are refused with a note naming the file, so nothing is ever silently left out of your map.

Map the key findings, method and limitations of this paper [attach PDF] Turn this scanned handout into a revision map [attach PDF] Map the architecture these files describe — one root per module [attach several source files] Structure this whiteboard photo into a proper map [attach image] Build a map comparing these two product sheets [attach 2 PDFs]

Chat with Your Map

The in-editor AI panel AI button

Inside any open map, tap the AI button (above the zoom controls) to open the chat panel. The assistant sees your map's structure and your selected node, so you can talk about it directly. On phones the panel is a compact bottom sheet — swipe down or tap the top arrow to dismiss it.

Direct commands act on the map — added nodes and edits are applied immediately and can always be undone.
Suggestions are propose-then-accept — the idea button (or asking for ideas) shows ghost nodes you can keep or drop, and reply options appear as tappable chips (select several, then confirm with the checkmark).
Animate connections Premium — ask to animate the link between two nodes and the assistant offers a few live animated previews right in the chat; tap one to apply it, or tap Other to describe exactly what you want (effects, an emoji to send along the line, direction and speed).
Attachments — the + button adds PDFs, images, text files and links to the conversation (see below).
Conversations — each map keeps its recent conversations; switch or start new ones from the title dropdown.

Working with attachments in chat

Attach with the + button, then say what to do with it. The assistant reads the attachment and edits your map from it — the same file types as map generation: PDFs (up to 2, 8 MB each), images, text and code files, and links.

Attachments stay available for follow-up questions. Attach a PDF once and you can keep asking about it for the rest of the conversation — including after closing and reopening the map — without attaching it again.

Build from a document

Add the key points of this PDF under “Research” [attach PDF] Now expand section 3 into its own branch What does the PDF say about cost? Add it as notes on the selected node

Compare against the map

Compare this spec to my map and add whatever’s missing [attach file] Which requirements in this document aren’t covered by the map yet?

From an image

What is in this screenshot? Map it. [attach image] Turn this diagram into nodes under the selected node [attach image]

If a file can’t be read (video, archive, Word or Excel), the panel says so straight away rather than letting you send a message about a document the assistant never received.

Command Reference — things to ask

Building & expanding

Add branches / nodes

Add a branch about competitor analysis with 5 sub-points Add three pros and three cons under the selected node

Expand with suggestions (ghost nodes)

Suggest ways to expand this map What am I missing under “Marketing”?

Enriching nodes

Notes & summaries

In chat the AI keeps nodes to short titles and adds notes only when you ask — request a description, summary, or details (or tell it to always add notes).

Write a short note on the selected node explaining the concept Summarize this whole branch into one note on its root From now on, always add a short note to every node you create

Links, tables, lists & tasks

Add a comparison table of the three options under “Vendors” Turn these nodes into a task checklist Add useful links about each framework

Pictures

Ask for images and you get the actual pictures placed on the nodes, not links to them. The assistant describes what each node should show and miMind looks the pictures up for you, so adding a dozen at once is quick.

Ask for an icon or a logo instead and you get line art on a transparent background, which sits cleanly on a coloured node. New maps get one of these on the central node automatically — a single image that sums up what the map is about.

Add a picture to each of the landmark nodes Put an image of the Eiffel Tower on the selected node Give the main node an icon for this topic

Videos

Ask for a video and the assistant attaches a player to the node. Every video is checked before it lands, so dead or private links don't end up on your map.

Add a good intro video to the selected node

Icons, emoji & visuals

Add fitting emoji to the top-level branches

Restructuring & understanding

Undoing what the AI did

Every reply that changed your map carries an Undo this change button, which reverses that whole turn in one press — however many nodes, pictures and edits it involved. You can also just type undo (or redo) in the chat box; those are handled instantly on your device and don't use any of your AI allowance.

undo redo

Update & retitle

Rename the selected node to something clearer Rewrite the titles in this branch to be shorter

Ask about the map

What does this map say about the timeline? Which branch needs the most work?

Work with attachments

Add the key points of this PDF under “Research” [attach PDF] Now expand section 3 — no need to attach the PDF again Compare this spec to my map and add what’s missing [attach file] What is in this screenshot? Map it. [attach image]

Use Your Own API Key

Bring-your-own Anthropic key

miMind AI subscribers with their own Anthropic account can use its token limits instead of miMind's. Open the main menu → Settings → AI Settings → AI API Key… and paste a key (starts with sk-ant-). The key is stored only on your device, sent only with your own requests, and cleared automatically when you sign out or your AI subscription ends.

Agentic AI — maps that act

Turn any node into a live agent Agentic AI

With the miMind Agentic AI tier, maps don't just plan — they act in the real world. Add an Agent from the node content menu (next to Location) and it becomes a live card inside your node with a glowing status halo: green healthy, yellow warnings or a pending question, red errors.

Directives — double-tap the agent card to set its name, directive prompt, and schedule: run manually, every 15 minutes / hour / 6 hours, daily, or on an external event via its personal webhook URL (works with Zapier, IFTTT, and any service that can POST).
Notifications & questions — agents report results to your inbox, and when they need a decision they ask with tappable answer options right on the card; your choice is remembered for future runs (adjustable in settings).
Context-aware — each agent knows the node it lives in, the content beside it, and where it sits in the larger map.
Connectors — wire named webhooks (services or devices: Zapier, IFTTT, Home Assistant, anything that accepts a POST) into an agent's settings; the agent can then trigger them by name — and only the ones you configured.
Undoable — every change an agent makes goes through miMind's normal editing pipeline.

Example directives

Every morning, check the tasks under this node and notify me about anything due this week When my store's webhook fires, summarize the order and add it under "Orders" Watch this reading list and suggest one new article a day

Plans & Limits

Plan	AI allowance	Notes
Free / Premium	5 AI requests per month	A taster of both map generation and chat. Resets on the 1st.
miMind AI	A generous monthly token budget with a daily burst cap	Roughly 100+ full map generations or several hundred chat turns per month. Includes the MCP server and the bring-your-own-key option. See Pricing.
miMind Agentic AI	Everything in miMind AI, plus a larger budget for agent runs	Live agents, directives, schedules, webhooks and notifications. See Pricing.

Your current usage is shown in Account → My Subscription. Fair use applies; abusive automation may be limited.

AI & Your Data

What’s sent, and what we keep

miMind’s AI features are powered by Anthropic (Claude). AI requests go through miMind’s secure gateway to Anthropic; the app never calls Anthropic directly.

Only when you use AI, and only what the request needs. Generating a map sends your topic, pasted text or URL; chatting sends your message plus the outline of the open map (node titles and structure) and which node you have selected; attachments send the images, PDFs, extracted file text, and links you add. When you supply a link, Claude may fetch that page and may search the web to answer.
We don’t store your prompts or the AI’s replies. Our gateway streams them to Anthropic and keeps only usage counts (tokens and request counts) for your plan’s limits — not the content. Agentic results delivered to your in-app inbox are the one exception, and you can delete them.
PDFs you attach in chat are held for up to 30 days. So that follow-up questions still work, a PDF attached in the chat panel is uploaded to Anthropic’s file storage and referenced by id on later turns, instead of being re-sent each time. We record only who owns each upload — never its contents — and a file can only ever be read back into your own conversations. Uploads are deleted automatically after 30 days. Attachments used for map generation are not stored at all: they travel with that one request.
Anthropic’s terms. We use Anthropic’s API under commercial terms under which your inputs and outputs are not used to train its models. See Anthropic’s Privacy Policy.
Your own key stays on your device. A bring-your-own Anthropic key is stored only on your device, used only for your requests, and cleared when you sign out or your AI subscription ends.
MCP sync is opt-in and publishes only map outlines — never your attachments or files; access tokens are stored only as hashes and are revocable at any time.
Availability. Because our AI provider does not serve certain sanctioned countries, AI features are unavailable there; core miMind features are not affected.

Full details are in the miMind Privacy Policy. Only submit content you’re comfortable processing with a third-party AI service.

The AI assistant built into your mind maps