# AI and Notion. URL: https://levystreet.com/integrations/notion/ Updated: 2026-10-19 How an AI Notion integration works in 2026: Notion AI plans, the API, webhooks and hosted MCP, with knowledge refresh and search designs for NZ teams. ## Short answer An AI Notion integration connects a model to your workspace through Notion AI, the REST API, webhooks or Notion's hosted MCP server, so it can find, summarise and maintain pages. For New Zealand teams the hard parts are not the model: they are which pages a connection can see, keeping answers current as pages change, and making AI suggest edits rather than rewrite the wiki. - Notion Agent, AI Meeting Notes and Enterprise Search start on Business: US$20 per member monthly, billed yearly. - Notion's hosted MCP server at mcp.notion.com uses OAuth and each user's own page access. - The API allows 180 requests a minute per connection, or 600 on Business and Enterprise. - Internal connections see only pages shared with them, so a shared bot shows everyone the same pages. - Content-update webhooks are batched and can arrive a minute or more after an edit. ## What Notion AI already does (October 2026) Notion's built-in AI covers most "ask the wiki" needs, if you pay for the right plan. On [Notion's pricing page](https://www.notion.com/pricing) as listed on 19 October 2026 (shown in USD to NZ visitors), Plus costs US$10 per member per month billed yearly and includes only a trial of Notion AI. Business, at US$20 billed yearly or US$24 monthly, adds Notion Agent, AI Meeting Notes and Enterprise Search. Custom Agents, which run repetitive tasks on their own, are free to try and then cost US$10 per 1,000 monthly Notion credits. A useful decision rule: if staff ask questions inside Notion and everyone is already on Business, start there. Build a custom integration when answers must also draw on other systems, appear in another channel, or write back through an approval step you control. ## Integration routes and when to use each | Route | What it gives you | Use it when | |---|---|---| | Notion AI and Custom Agents | Search, Q&A, meeting notes and agents inside Notion, billed by plan and credits | Work happens in Notion and Business pricing is acceptable | | REST API | Pages, data sources, comments and users through a bearer token | You run a scheduled sync, a search index or a controlled write workflow | | Webhooks | Events such as `page.content_updated` and `comment.created`, signed with an `X-Notion-Signature` HMAC | You need to re-index when pages change instead of polling | | Hosted MCP server | Search, fetch, create, update and trash pages from Claude, ChatGPT and other MCP clients, as the signed-in user | Individuals want Notion inside their assistant; see our [MCP guide](/engineering/mcp-for-business-systems/) | | n8n, Make or Zapier | Prebuilt Notion steps plus raw HTTP calls | Low-volume flows such as "form entry creates a page" | ## Automation 1: keep the knowledge base current Stale pages are the main reason AI answers from a wiki go wrong. This job finds them and asks the owner to act; it never edits page bodies. - **Inputs**: Pages in a "Policies and procedures" data source with Owner, Status and Last reviewed properties, plus newer pages and comments on the same topic found through search. - **Model step**: For each page not reviewed in 180 days, list what it claims, compare with newer material, and name specific conflicts with links to both sources. - **Human approval**: The page owner receives a comment listing conflicts and a suggested correction, then accepts, edits or dismisses it. - **Write-back**: Comment through the API; set Status to "Needs review". Last reviewed changes only when the owner signs off. - **Failure and recovery**: Pages the run did not reach keep their old Status, so a filtered view shows what to rerun. Pages moved to Trash through MCP can be restored with `notion-restore-pages` or from Trash. ## Automation 2: permission-aware answers from Notion The [authorization docs](https://developers.notion.com/docs/authorization) set the constraint: an internal connection can read a page only after it is manually shared with that connection, and sharing a parent page shares its children. A Teams or website bot built on one internal connection therefore shows every user everything the connection can see. 1. **Pick the access model.** For personal assistants, use the hosted MCP server or a public OAuth connection, so each person reads only what they can read. For a shared bot, create one connection per audience (all staff, managers) and share only that audience's pages with it. 2. **Index with metadata.** Store page ID, URL, teamspace and last edited time with every chunk, and re-index on webhook events. Our guide to [building private AI search](/engineering/building-private-ai-search/) covers the indexing design. 3. **Cite or decline.** Every answer links the pages it used. If retrieval finds nothing relevant, the bot says so rather than guessing. 4. **Check before relying on a result.** Notion notes that AI search results omit a page's verification state, so fetch the page itself when that matters. ## Permissions and approval design Give back-office connections read and comment capabilities only, and add write access per workflow rather than by default. On the MCP side, workspace-owned connections expose only the tools an admin selects, MCP tokens can be created only by workspace owners on Business and Enterprise, and limiting which clients members connect needs Enterprise, per [Notion's MCP docs](https://developers.notion.com/docs/mcp). Do not rely on an MCP client to ask before it writes; put approvals in your own layer, as in our [RAG development](/services/rag-development/) and [AI integration](/services/ai-integration/) work. Wiki pages often hold staff and customer personal information. Under the Privacy Act 2020 you remain responsible for it when a provider processes it on your behalf, so record which model providers see Notion content and their retention terms. ## Costs and limits (as listed 19 October 2026) - **API**: 180 requests per minute per connection, or 600 on Business and Enterprise, plus a shared per-workspace limit. `Retry-After` is at most 60 seconds. - **Free workspaces**: those with more than one member are capped at 1,000 lifetime blocks through the API, effective 8 September 2026. - **Model cost, worked example**: a weekly refresh over 400 pages at about 3,000 input and 300 output tokens each, on Claude Sonnet 5.5 at US$2 and US$10 per million tokens, is roughly 400 × US$0.009 = US$3.60 a run. ## Where it goes wrong - **Late events.** `page.content_updated` is aggregated and may arrive a minute or two after an edit. Add a nightly pass on last edited time to catch anything missed. - **Partial pages.** Very large pages can come back from `notion-fetch` with `truncated` set. Check the flag before letting the model answer from them. - **Rate-limit errors.** Notion MCP retries once when the wait is two seconds or less; otherwise it returns a `rate_limited` tool error. Agents that fan out in parallel hit this first. - **Instructions inside pages.** Pages edited by guests or pasted from email can contain text aimed at the model. Treat page content as data; see [recognising prompt injection](/learn/recognising-prompt-injection/). Teams holding documents in more than one place should read our [SharePoint](/integrations/sharepoint/) and [Google Drive](/integrations/google-drive/) guides, since cross-system search changes the permission design. ## Questions people ask ### Does Notion AI train on our workspace data? Notion's pricing page says Notion AI will not use your data to train its models unless you opt in to a request to share data, and the Enterprise plan adds zero data retention with its LLM providers. Your own integrations are separate: if you send Notion content to OpenAI, Anthropic or Google through their APIs, check that provider's retention and training terms as well. ### Can ChatGPT or Claude connect to Notion directly? Yes. Notion runs a hosted MCP server at https://mcp.notion.com/mcp that MCP clients such as Claude and ChatGPT connect to with OAuth. Each person signs in with their own Notion account, so the assistant can only reach pages they can reach. Admins can limit which tools a workspace-owned connection exposes, and restricting which clients members connect needs the Enterprise plan. ### Which Notion plan do we need for AI features? As listed on 19 October 2026, the full Notion AI set (Notion Agent, AI Meeting Notes and Enterprise Search) starts on Business at US$20 per member per month billed yearly, or US$24 monthly. Free and Plus get a trial of Notion AI only. Custom Agents are free to try, then cost US$10 per 1,000 monthly Notion credits. ### Should we use Notion AI or build our own search over Notion? Use Notion AI when staff ask questions inside Notion and everyone is on Business anyway. Build your own when answers must combine Notion with other systems such as SharePoint or a CRM, appear in another channel such as Teams or a website, follow your own retention and logging rules, or write back through an approval step you control. ## Sources - [Notion pricing](https://www.notion.com/pricing) - [Request limits](https://developers.notion.com/reference/request-limits) - [Workspace block limits](https://developers.notion.com/reference/workspace-block-limits) - [Notion MCP](https://developers.notion.com/docs/mcp) - [Notion MCP supported tools](https://developers.notion.com/docs/mcp-supported-tools) - [Webhooks](https://developers.notion.com/reference/webhooks) - [Authorization](https://developers.notion.com/docs/authorization) - [Pricing](https://platform.claude.com/docs/en/about-claude/pricing) - [Working with third-party providers](https://www.privacy.org.nz/resources-and-learning/a-z-topics/working-with-third-party-providers/) Published by Levy Street (https://levystreet.com), Wellington, New Zealand.