CtrlK
Notion AI is a suite of artificial intelligence features integrated into Notion, a widely used productivity and collaboration platform. First released in private alpha on November 16, 2022, Notion AI became generally available on February 22, 2023. The AI capabilities span writing assistance, question-and-answer search across workspaces, database autofill, summarization, translation, and (since September 2025) autonomous AI agents. Notion AI is powered by multiple large language models from OpenAI and Anthropic, and serves over 100 million Notion users worldwide.
Notion Labs, Inc. was founded in 2013 in San Francisco, California. The five co-founders were Ivan Zhao (CEO), Simon Last (CTO), Chris Prucha, Jessica Lam, and Toby Schachman. The original concept positioned Notion as a no-code website builder for non-technical users, allowing people to create web pages and simple applications without programming knowledge.
The company raised approximately $2 million in seed funding from angel investors in March 2013. However, by 2015 the startup had nearly run out of money, and Zhao and Last faced a difficult choice. They decided to lay off their small team, leave San Francisco, and relocate to Kyoto, Japan, where they rebuilt Notion from scratch with a new technical architecture and a broader product vision.
In August 2016, the team shipped Notion 1.0, an all-in-one workspace combining notes, tasks, wikis, and databases. Notion 2.0 followed in March 2018, introducing a more flexible block-based editor that let users arrange content with drag-and-drop blocks representing text, images, tables, embeds, and more.
Notion's user base grew rapidly after the 2.0 release. By September 2019, the company had reached 1 million users. Growth accelerated further during the COVID-19 pandemic as remote teams adopted Notion for distributed collaboration.
| Funding Round | Date | Amount Raised | Valuation | Lead Investor(s) |
|---|---|---|---|---|
| Seed | March 2013 | ~$2 million | N/A | Angel investors |
| Series A | April 2019 | $10 million | $800 million | Sequoia Capital |
| Series B | July 2020 | $50 million | $2 billion | Index Ventures |
| Series C | October 2021 | $275 million | $10 billion | Coatue Management, Sequoia Capital |
By October 2021, when the company closed its $275 million Series C round at a $10 billion valuation, Notion reported 20 million users. As of December 2025, the company's valuation stands at approximately $11 billion, and its annual revenue reached an estimated $600 million, up from roughly $400 million in 2024.
In September 2024, Notion announced that it had surpassed 100 million users globally, a milestone Ivan Zhao highlighted in a blog post. Roughly 80% of Notion's user base is located outside the United States.
Notion AI was announced and released in private alpha on November 16, 2022. The feature was initially described as a writing assistant powered by GPT-3-era technology from OpenAI. Notion opened a waitlist for early access, expecting perhaps 200,000 sign-ups. The actual response far exceeded expectations, with over 2 million users joining the waitlist.
During the 10-week alpha period, Notion collected feedback from early testers and iterated on the product. Engineers and product managers used the data to refine the AI's output quality, identify the most popular use cases, and prepare for a broader rollout.
On February 22, 2023, Notion AI launched to all users. The initial feature set focused on inline writing assistance: users could invoke AI from any text block in Notion to draft content, improve existing writing, fix spelling and grammar, change tone, make text shorter or longer, summarize pages, translate between languages, and brainstorm ideas.
At launch, Notion offered 20 free AI responses per workspace member. After this trial allocation (which expired on April 5, 2023), continued use required purchasing the Notion AI add-on at $10 per member per month ($8 per member per month with annual billing).
The writing assistant is the foundational Notion AI feature and remains the most frequently used capability. It operates directly within Notion's block-based editor, meaning users do not need to leave their workspace or switch to a separate application.
| Feature | Description |
|---|---|
| Draft with AI | Generate first drafts of blog posts, meeting agendas, press releases, product specs, emails, and other document types from a short prompt. |
| Improve writing | Rewrite selected text for clarity, flow, and readability. |
| Fix spelling and grammar | Detect and correct typographical and grammatical errors in existing text. |
| Change tone | Adjust the register of a passage to be more professional, casual, friendly, or direct. |
| Make shorter / Make longer | Compress or expand selected content while preserving the original meaning. |
| Summarize | Condense a page or selection into a brief summary capturing the main points. |
| Translate | Translate text into dozens of supported languages without leaving the page. |
| Brainstorm ideas | Generate lists of ideas, names, concepts, or outlines based on a topic or prompt. |
| Explain | Simplify technical or complex passages into plain language. |
| Extract action items | Pull out tasks and next steps from meeting notes or lengthy documents. |
All of these actions are available through an inline AI menu that appears when a user types /ai or selects a block and chooses an AI option. The assistant is context-aware, meaning it can read the surrounding content on the page to produce more relevant outputs.
On November 14, 2023, Notion introduced Q&A, a conversational search feature that answers natural-language questions by drawing on information stored across a user's entire Notion workspace. Unlike traditional keyword search, Q&A understands the intent behind a question and synthesizes an answer from multiple pages, databases, and documents.
Q&A respects Notion's existing permission model: it only references pages the asking user has access to view. Each answer includes citations linking back to the source pages, so users can verify the information and read further context.
Typical Q&A use cases include:
Q&A is accessible through a sparkle icon in the bottom-right corner of the Notion interface, through the search menu, or via the keyboard shortcut. When AI connectors are enabled, Q&A can also pull information from connected external tools such as Slack, Google Drive, and Jira.
Introduced in mid-2023, the autofill feature brings AI capabilities directly into Notion databases. Users can add AI-powered properties (columns) to any database, and Notion AI will automatically populate them based on the content of each database page.
Notion provides several pre-built AI property types:
| AI Property Type | Function |
|---|---|
| AI Summary | Generates a concise summary of the page content. |
| AI Key Info | Extracts important details such as people mentioned, dates, action items, and decisions. |
| AI Translation | Translates the page content into a specified target language. |
| Custom Autofill | Runs a user-defined prompt against the page content, enabling highly specific extractions or transformations. |
The custom autofill property is particularly flexible. Users write their own prompt (for example, "Extract the product name and pricing from this page" or "Generate an SEO meta description"), and Notion AI executes it for every row in the database. An auto-update option can be enabled so that the AI property refreshes automatically five minutes after any edit to the underlying page.
This feature is widely used for managing content calendars, CRM-style databases, product backlogs, and research libraries, where manual tagging and summarization would otherwise require significant effort.
Announced in stages during 2024, Notion AI connectors allow the AI to search and retrieve information from external applications that a team uses alongside Notion. Connectors were formally introduced as part of the "new Notion AI" release on September 25, 2024.
As of early 2026, the supported connectors include:
| Connector | Content Indexed |
|---|---|
| Slack | Public channel messages (not private channels or DMs) |
| Google Drive | Google Docs, Sheets, Slides, and other Drive files |
| Jira | Jira Cloud projects, issues, and comments |
| GitHub | Repositories, issues, and pull requests |
| Microsoft Teams and SharePoint | Messages and shared documents |
| Linear | Issues, projects, and comments |
| OneDrive | Files and documents |
| Gmail | Email messages (rolling out in 2025) |
| Box | Files and documents |
When a connector is activated, Notion begins indexing the connected app's content, a process that can take up to 72 hours for the initial sync. Once indexed, Q&A and other AI features can reference information from these external sources, citing the specific Slack message, Google Doc, or Jira issue that informed the answer.
Connectors are available on Notion's Business and Enterprise plans. The integration can access information dating back approximately one year from the date the connector is set up.
Launched on May 13, 2025 as part of Notion 2.51, Enterprise Search extends AI-powered search beyond Notion's own pages to encompass a wide range of connected tools. When a user searches, Notion AI looks across Notion pages, connected Microsoft tools (Teams, SharePoint, OneDrive), Gmail, Linear, and other integrated apps, returning summarized answers rather than just a list of links.
Enterprise Search also indexes all PDFs uploaded to Notion workspaces, making previously unsearchable document content available through the AI.
Also introduced with Notion 2.51 in May 2025, Research Mode is designed for complex, multi-source questions. When activated, it performs a deep investigation across the user's entire Notion workspace, connected apps (Slack, Google Drive, and others), and the public internet. Rather than returning a single short answer, Research Mode compiles a structured report with cited sources, analysis, and synthesized findings.
Released alongside Enterprise Search in May 2025, AI Meeting Notes is a new block type that automatically transcribes and summarizes conversations. When integrated with Notion Calendar, AI meeting notes can be added to every meeting automatically, capturing key discussion points, decisions, and action items without requiring manual note-taking.
Notion Mail launched on April 15, 2025 as an AI-powered email client that connects to Gmail accounts. It uses Notion AI to help users organize their inbox, draft replies, adjust tone, and schedule meetings. Key features include AI auto-labeling (which sorts emails into topic-specific views), smart scheduling (which checks the user's calendar and suggests available times), and AI-assisted composition. Notion Mail is available on macOS, Windows, and the web.
On September 18, 2025, Notion released version 3.0, a major update that introduced AI agents. These agents can perform any action a user can take in Notion, but at much greater scale and over extended time periods (supporting over 20 minutes of continuous multi-step work).
Personal agents can handle entire projects: building launch plans, breaking them into tasks, assigning work to team members, and drafting initial documents. In databases, agents can update or create hundreds of pages simultaneously. A custom instructions page allows each user to configure how their agent behaves, and a memory system helps agents retain context across interactions.
On February 24, 2026, Notion shipped Custom Agents, which are fully autonomous agents that operate on schedules or triggers without requiring manual prompting. Users define a "job description" for each agent, set a trigger condition or recurring schedule, and the agent executes the work independently, around the clock.
Use cases for custom agents include:
Custom agents also integrate with external tools such as Slack and Figma. During the early access period, testers created over 21,000 custom agents. Notion itself reported running 2,800 agents internally at the time of launch.
Notion offered custom agents at no additional cost through May 3, 2026. After that date, a new credit-based pricing model applies at $10 per 1,000 Notion Credits.
Notion AI does not rely on a single language model. Instead, it uses a multi-model architecture, routing different tasks to different models depending on the requirements.
When Notion AI launched in late 2022, it initially used OpenAI's models. According to Notion's own account, the company received early access to GPT-4 in the fall of 2022 and built its AI features on top of that capability. The writing assistant and early Q&A features were powered primarily by OpenAI's GPT series.
Over time, Notion expanded its model portfolio to include models from Anthropic. The Notion engineering team found that Claude's reasoning and instruction-following capabilities, along with its conversational tone, were well-suited to writing and editing tasks. Notion described Claude as "the clear leader across all criteria" for its one-click writing and editing skills.
In May 2025, Notion formally announced support for model switching within the workspace, enabling users to chat with different LLMs and choose the model best suited to their task. The initially supported models were OpenAI's GPT-4.1 and Anthropic's Claude 3.7 Sonnet.
By September 2025 (with Notion 3.0), the platform offered access to newer models including Claude Sonnet 4 and GPT-5 from OpenAI. Users on Business and Enterprise plans can toggle between available models based on their specific needs.
| Model Provider | Models Used | Primary Use Cases |
|---|---|---|
| OpenAI | GPT-4.1, GPT-5, o3, o1-mini | Structured content generation, reasoning tasks, general chat |
| Anthropic | Claude 3.7 Sonnet, Claude Sonnet 4 | Writing, editing, tone adjustment, instruction-following |
In addition to these third-party models, Notion has developed its own internal models and fine-tuned versions optimized for Notion-specific retrieval tasks. These custom models help Notion AI understand workspace context, business terminology, and user intent with lower latency.
Notion also uses an embeddings pipeline: each Notion page is converted into a vector embedding and stored in a vector database (Turbopuffer), which enables fast semantic search and retrieval for Q&A and Enterprise Search features.
Notion AI pricing has evolved significantly since launch.
When Notion AI became generally available in February 2023, it was offered as an add-on to any Notion plan at $10 per member per month ($8 per member per month with annual billing). All Notion plan tiers (Free, Plus, Business, and Enterprise) could purchase the add-on.
In 2025, Notion eliminated the separate AI add-on. Full AI capabilities, including Q&A, connectors, Research Mode, Enterprise Search, AI Meeting Notes, and agents, were bundled into the Business and Enterprise plans. The change took effect on August 13, 2025.
| Plan | Monthly Price (per user) | Annual Price (per user/month) | AI Access |
|---|---|---|---|
| Free | $0 | $0 | Basic AI writing (limited) |
| Plus | $12 | $10 | Basic AI writing features |
| Business | $24 | $20 | Full AI: agents, Q&A, connectors, Research Mode, Enterprise Search, model switching |
| Enterprise | Custom | Custom | Full AI plus advanced admin controls, SSO, SCIM, audit logs, zero data retention with LLM providers |
Users who had previously purchased the AI add-on on Free or Plus plans retained access to their existing features even after the pricing change. However, new users on Free and Plus plans can no longer purchase the add-on.
For custom agents (introduced February 2026), Notion adopted a credit-based model: $10 per 1,000 Notion Credits, taking effect May 4, 2026.
Notion has published detailed documentation about its AI security practices, and the AI features are included within the scope of the company's SOC 2 Type 2 report and ISO 27001 certification. Notion holds four ISO certifications: ISO 27001, ISO 27701, ISO 27017, and ISO 27018.
Key privacy commitments include:
Enterprise security features include SSO via SAML 2.0, user provisioning through SCIM, audit log tracking, and granular permission controls.
Notion AI operates in a competitive market alongside several major AI-powered productivity platforms.
Microsoft 365 Copilot integrates AI across Word, Excel, PowerPoint, Outlook, and Teams. It is powered primarily by OpenAI's GPT-4 series and is priced at approximately $30 per user per month on top of existing Microsoft 365 licenses. Copilot is targeted at large enterprises already embedded in the Microsoft ecosystem and offers deep integration with Office applications.
Compared to Notion AI, Microsoft Copilot provides stronger spreadsheet and presentation capabilities (through Excel and PowerPoint integration) and more mature enterprise administration tools. However, Notion AI benefits from its unified workspace model, where documents, databases, wikis, and projects all live in a single connected environment rather than across separate applications.
Google has integrated its Gemini AI models into Google Docs, Sheets, Slides, Gmail, and Drive. The Gemini features include "Help me write" in Docs, "Help me organize" in Sheets, image generation in Slides, and "Ask Gemini in Drive" for cross-document search.
Google Workspace AI is cost-effective for organizations already using Google's suite and offers strong collaborative editing features. However, because Google Workspace consists of separate applications (Docs, Sheets, Drive, Gmail, Chat) that share a login but lack a unified data model, the AI must work across application boundaries. Notion's architecture, built around a single connected workspace with relational databases, gives its AI a structural advantage when answering questions that span multiple types of content.
Other notable competitors in the AI-powered productivity space include:
| Competitor | Key AI Features |
|---|---|
| Coda AI | AI writing, formula assistant, and automations within Coda documents |
| Confluence (Atlassian Intelligence) | AI summarization, writing, and search across Confluence and Jira |
| Clickup AI | AI writing, summarization, and task management features within ClickUp |
| Mem | AI-first note-taking with automatic organization and retrieval |
| Craft | AI writing assistant integrated into the Craft document editor |
| Date | Event |
|---|---|
| 2013 | Notion Labs founded in San Francisco by Ivan Zhao, Simon Last, Chris Prucha, Jessica Lam, and Toby Schachman. |
| August 2016 | Notion 1.0 released. |
| March 2018 | Notion 2.0 released with block-based editor. |
| September 2019 | Notion reaches 1 million users. |
| October 2021 | Series C funding at $10 billion valuation; 20 million users. |
| November 16, 2022 | Notion AI private alpha launched; over 2 million waitlist sign-ups. |
| February 22, 2023 | Notion AI general availability. |
| Mid-2023 | AI autofill properties for databases introduced. |
| November 14, 2023 | Q&A by Notion AI launched in beta. |
| June 2024 | Slack connector and GPT-4 integration announced. |
| September 2024 | 100 million users announced; AI connectors (Slack, Google Drive, and more) formally introduced. |
| April 15, 2025 | Notion Mail launched. |
| May 13, 2025 | Notion 2.51: AI Meeting Notes, Enterprise Search, Research Mode, and model switching (GPT-4.1, Claude 3.7). |
| August 13, 2025 | AI add-on eliminated; AI bundled into Business and Enterprise plans. |
| September 18, 2025 | Notion 3.0: AI agents with multi-step actions and memory. |
| February 24, 2026 | Notion 3.3: Custom agents with autonomous operation on schedules and triggers. |