What you'll be able to do
- Draft a doc, then tighten its tone, length, and structure with AI in place
- Summarize a wall of notes into decisions and clear action items
- Autofill a database property — summaries, categories, next steps — across every row
- Ask a question and get an answer sourced from your own pages
- Build a wiki that stays useful instead of going stale
Inside the path
A focused set of five-minute lessons — each one ends with a hands-on exercise, not a quiz you can guess.
Draft without the blank page 5 min
Prompt Notion AI for a first draft, then edit it into your voice right in the doc.
Notes to action items 5 min
Turn raw meeting notes into a checklist with owners and due dates you can assign.
Summaries people actually read 5 min
Get a tight summary at the top of a long doc, aimed at one specific reader.
AI inside databases 6 min
Use AI autofill to categorize, summarize, or extract a field across every row.
Ask your workspace 5 min
Use Q&A to find and synthesize answers from your own pages, with links to the sources.
Try a sample exercise
This is the kind of card you'd practice inside Iro — you do the thinking, then get feedback.
◆ Sample exercise · Prompt practice
You paste 20 lines of messy meeting notes into a Notion doc and want a clean list of next steps your team can actually act on.
Your task: Choose the Notion AI prompt that turns the notes into something usable.
- "Summarize these notes."
- "From the notes above, extract every action item as a checklist. For each one, name the owner mentioned and any due date, and mark items with no owner as 'Unassigned'. Ignore general discussion that isn't a decision or a task."
- "Make this look better."
- "What should we do next?"
See why the second prompt wins
The winning prompt is grounded in the notes above and asks for a specific output format (a checklist), names the exact fields to extract (owner and due date), handles an edge case (mark items with no owner as 'Unassigned'), and sets the scope (ignore general discussion). "Summarize these notes" gives you a paragraph you still have to read; "what should we do next?" asks the AI to decide for you rather than pull what your team already agreed. In Iro you'd write your own version and get feedback on the fields and edge cases you missed.
Notion AI is context-hungry — feed it your pages
Unlike a blank chatbot in another tab, Notion AI lives inside your content. It can read the doc you're in, pull from a database, and with Q&A search across your entire workspace (and connected tools like Slack or Google Drive on supported plans) to answer with links to the source pages. That means the quality of what you get out is tied to what you put in: a well-kept workspace makes Notion AI genuinely smart about your own work.
To trigger it, press space on an empty line or highlight text and choose Ask AI, then either pick an action (draft, summarize, improve, translate) or type your own instruction. As with any AI, the more precise you are about the format — a checklist, a three-bullet summary, a specific database field — the less editing you do afterward.
The prompting skill isn't Notion-specific
Extracting structured output from messy input — action items with owners, a summary for a named reader, a category for each row — is the same skill you'd use in ChatGPT, Claude, or Copilot. What is special about Notion AI is that your content is already there, so you don't paste anything, and Q&A can cite the exact page an answer came from. Learn the extract-to-format habit here and it pays off in every AI tool you touch.