Notion AI Prompts

Act as my research analyst inside Notion. I need a digest on [RESEARCH_TOPIC] drawn only from my workspace: the databases [DATABASE_1] and [DATABASE_2], and the pages under [PARENT_PAGE]. Ask Notion AI these questions in order: first, 'What do my notes in [DATABASE_1] say about [SUB_QUESTION_1]?' then 'Summarize every mention of [KEYWORD] across [PARENT_PAGE] and its subpages,' then 'What decisions did we record about [DECISION_TOPIC] and when?' Compile the answers into a new page titled '[DIGEST_TITLE]' with sections: Key findings ([FINDING_COUNT] bullets), Open questions ([QUESTION_COUNT] bullets), and Source links (every claim linked to its source page). Add the constraint: if the workspace has no information on [GAP_TOPIC], say so explicitly instead of guessing. Output format: the three questions to paste into Notion AI Q&A, plus the digest page structure.

Act as a CRM administrator. My Notion database '[DATABASE_NAME]' has [ROW_COUNT] company rows with properties: [PROPERTY_1] (text), [PROPERTY_2] (select), [PROPERTY_3] (date), and [PROPERTY_4] (URL), but many are empty. The page content of each row contains [CONTENT_DESCRIPTION] (meeting notes, emails, research). Use Notion AI Autofill to populate the missing properties: configure the prompt for [PROPERTY_1] as 'Extract the [FIELD_DETAIL] mentioned in this page, output only the value,' for [PROPERTY_2] choose strictly from the options [OPTION_LIST], for [PROPERTY_3] output in [DATE_FORMAT], and skip rows where the page lacks the information instead of inventing it. Run autofill on a test batch of [TEST_BATCH_SIZE] rows, verify [VERIFICATION_SAMPLE] of them manually, then run the full database. Output format: the exact autofill prompt per property, the test plan, and the verification checklist.

Act as my chief of staff. I will paste the raw transcript or my rough notes from [MEETING_TYPE] about [MEETING_TOPIC] with attendees [ATTENDEE_1], [ATTENDEE_2], and [ATTENDEE_3] into a Notion page. Instruct Notion AI: 'Turn these notes into structured meeting notes with four sections: 1) Decisions made ([DECISION_COUNT] max, each with owner), 2) Action items as a table with columns Task, Owner, Due date ([DUE_DATE_FORMAT]), 3) Open questions, 4) A [SUMMARY_LENGTH] sentence summary for [STAKEHOLDER]. Then change the tone of the summary to [TONE] for sharing with [AUDIENCE].' Add the rule: every action item must have a named owner from the attendee list or be flagged as unassigned; never assign to [NEVER_ASSIGN]. Output format: the exact instruction block to paste into Notion AI, plus the table column definitions.

Act as my writing partner inside Notion. I have a draft of [ARTICLE_TITLE], a [CONTENT_TYPE] (blog post, newsletter, guide) about [TOPIC] for [TARGET_AUDIENCE], currently [CURRENT_WORD_COUNT] words ending mid-section at [SECTION_NAME]. Instruct Notion AI: 'Continue writing from where the draft ends, keeping my voice: [VOICE_DESCRIPTION] (e.g. direct, warm, no jargon). Add [SECTIONS_TO_ADD] new sections covering [NEW_SECTION_TOPIC_1] and [NEW_SECTION_TOPIC_2], each [PARAGRAPHS_PER_SECTION] paragraphs. Then improve writing on the full draft for clarity, shorten any sentence over [MAX_WORDS_PER_SENTENCE] words, and translate the final call to action into [SECOND_LANGUAGE].' Set the constraint: do not change my stated facts about [FACT_TO_PRESERVE]; flag anything you are unsure about with [FLAG_MARKER]. Output format: the exact multi-step instruction to paste into Notion AI.

Act as an executive summarizer. My Notion page '[PAGE_TITLE]' documents [PROJECT_NAME], a [PROJECT_TYPE] running since [START_DATE], and it has grown to [APPROX_WORD_COUNT] words of notes, decisions, and threads. Instruct Notion AI: 'Summarize this page for [READER_ROLE] (e.g. a new team member, my manager) in [SUMMARY_FORMAT]: a [BULLET_COUNT] bullet executive summary, then a table of key decisions (Decision, Date, Owner), then current status of [STATUS_AREAS] (timeline, budget, risks). Keep it under [MAX_WORDS] words and link each decision to its source block.' Then ask a follow-up: 'What are the top [RISK_COUNT] risks or unresolved items on this page?' Output format: the two instructions to paste into Notion AI, in order.

Act as my localization editor in Notion. I have a [DOC_TYPE] titled '[DOC_TITLE]' written in [SOURCE_LANGUAGE] with a [CURRENT_TONE] tone, and I need versions for [AUDIENCE_1] in [TARGET_LANGUAGE_1] and [AUDIENCE_2] in [TARGET_LANGUAGE_2]. Instruct Notion AI in two steps: step 1, 'Translate this document into [TARGET_LANGUAGE_1], keeping formatting, headings, and the terms [TERMS_TO_KEEP] untranslated.' Step 2, on the translated copy, 'Change the tone to [NEW_TONE_1] for [AUDIENCE_1] (e.g. formal for enterprise clients) and to [NEW_TONE_2] for [AUDIENCE_2] (e.g. friendly for community members), without changing any facts, numbers, or dates.' Add the QA rule: back-translate [QA_SAMPLE] key sentences to check meaning survived. Output format: the two-step instruction block plus the QA checklist.

Act as a project coordinator. My updates on [PROJECT_NAME] are scattered across [NOTE_COUNT] Notion pages under [PARENT_PAGE], covering the period [DATE_RANGE]. Instruct Notion AI: 'Read all pages under [PARENT_PAGE] and generate a status table with columns: Workstream ([WORKSTREAMS]), Status (On track / At risk / Blocked), Owner, Last update date, Next milestone. Then below the table, write a [PARAGRAPH_COUNT] paragraph narrative summary for [STAKEHOLDER_AUDIENCE], and list anything dated before [STALE_CUTOFF] as possibly stale.' Set the rule: mark any workstream with no update since [STALE_CUTOFF] as 'Unknown, needs owner input' rather than guessing its status. Output format: the exact instruction to paste into Notion AI plus the table column definitions.

Act as my content strategist in Notion. I publish [CONTENT_TYPE] about [NICHE] for [TARGET_AUDIENCE] on [PLATFORM], posting [FREQUENCY]. Instruct Notion AI: 'Brainstorm [IDEA_COUNT] ideas for a series called [SERIES_NAME] about [SERIES_TOPIC]. For each idea give: working title, one-line hook, and which audience pain point ([PAIN_1], [PAIN_2], [PAIN_3]) it addresses. Then take the best [SHORTLIST_COUNT] and expand each into an outline with [OUTLINE_SECTIONS] sections, and draft the opening [OPENING_WORD_COUNT] words in my voice: [VOICE_DESCRIPTION].' Add the constraint: every idea must be answerable from my existing workspace knowledge about [KNOWLEDGE_AREA]; flag any idea needing outside research with [FLAG]. Output format: the instruction block plus the outline template.

Act as my productivity coach. I keep daily notes in the Notion database '[DAILY_NOTES_DB]' with entries for [WEEK_DATE_RANGE]. Instruct Notion AI: 'Read this week's daily notes and produce a weekly review page titled [REVIEW_TITLE] with: 1) Wins ([WIN_COUNT] bullets, tied to my goals [GOAL_1] and [GOAL_2]), 2) What did not get done and why ([MISSED_COUNT] items max), 3) Time drains spotted ([DRAIN_EXAMPLES]), 4) Next week's top [PRIORITY_COUNT] priorities drafted as tasks with suggested owners and due dates. Change the tone to [TONE] (honest but encouraging).' Add the rule: base every claim on the daily notes; if a goal has no supporting entries, write 'no evidence this week' instead of inventing progress. Output format: the instruction block plus the review page template.

Act as a support lead. My Notion workspace has [PAGE_COUNT] support and process pages under [SUPPORT_PARENT_PAGE] for [PRODUCT_OR_SERVICE], used by [TEAM_OR_CUSTOMERS]. Instruct Notion AI: 'Read every page under [SUPPORT_PARENT_PAGE] and draft an FAQ page titled [FAQ_TITLE] with [QUESTION_COUNT] questions, grouped under [CATEGORY_1], [CATEGORY_2], and [CATEGORY_3]. Each answer must be under [MAX_WORDS_PER_ANSWER] words, link to its source page, and use the tone [TONE] for [READER_AUDIENCE]. Then list [GAP_COUNT] questions the workspace cannot answer, so I know what documentation to write next.' Add the constraint: never invent policies not stated in the source pages; mark uncertain answers with [UNCERTAIN_MARKER]. Output format: the instruction block plus the FAQ structure.