🗾ITmedia AI+ (日本)•Stalecollected in 83m
Osaka Gas Cuts 2000 Hours Using Notion AI
💡A concrete enterprise case study on how to reclaim thousands of hours using AI-powered knowledge management.
⚡ 30-Second TL;DR
What Changed
Reduced 2,000 hours of monthly manual information search
Why It Matters
This case study proves that AI-powered knowledge management is a high-ROI application for large enterprises. It provides a blueprint for companies struggling with internal information fragmentation.
What To Do Next
Implement a RAG-based search system on your internal documentation to replicate these efficiency gains.
Who should care:Enterprise & Security Teams
🧠 Deep Insight
Web-grounded analysis with 19 cited sources.
🔑 Enhanced Key Takeaways
- •Osaka Gas's implementation of Notion AI involved a significant focus on operational design, including defining data storage locations, structuring rules for content creation, and separating roles for chat and file management to optimize for AI-driven data utilization.
- •Notion AI's knowledge hub leverages Retrieval-Augmented Generation (RAG) technology, which ensures that generated answers are directly derived from the company's internal data, thereby enhancing accuracy and preventing the creation of fabricated information.
- •The platform's evolution, particularly with Notion 3.0 (September 2025) and 3.2 (January 2026), transformed Notion AI from a simple assistant into an 'agent that executes' multi-step workflows, capable of autonomous operation for up to 20 minutes and integrating with external tools like Slack and Google Drive for comprehensive enterprise search.
- •Osaka Gas's strategic approach views generative AI as a 'capable subordinate,' aiming to redefine human-AI collaboration by shifting from human-centric data search to AI-centric data summarization and decision support, allowing employees to focus on higher-value tasks.
📊 Competitor Analysis▸ Show
| Feature / Product | Notion AI | ClickUp Brain | Coda AI | Coworker AI | Taskade Genesis |
|---|---|---|---|---|---|
| Primary Focus | Knowledge management, content, meetings, AI agents within Notion workspace | Structured project & task management | Formula-driven docs, data-driven teams, apps | Enterprise AI platform across full tool stack (cross-tool agents, org memory) | AI-native workspace, custom AI agents, workflow automation |
| AI Architecture | Agent-first (20-min autonomy), RAG, choice of GPT-5.2, Claude Opus 4.5, Gemini 3 | AI writing assistance, task automation | AI for writing & data, formula-based automation | Cross-org knowledge graph (OM1), autonomous agents | Workspace DNA (Memory, Intelligence, Execution loop) |
| Cross-Tool Integration | Searches across Slack, Google Drive, GitHub, Jira (beta), Microsoft Teams, Box, OneDrive, Salesforce, Asana (beta) | Primarily within ClickUp, AI writing assistance | 300+ native integrations | 50+ integrations across entire stack (Salesforce, Jira, Slack, email, meetings) | 100+ integrations |
| Pricing (per user/month, annual) | Business plan: $20 (includes full AI access) or AI add-on: $8-$10 (on top of other plans) | AI add-on: $5 (on top of paid plans) | Included in plans (e.g., Pro $10, Team $30) | $30 | Free plan includes basic agents; Starter $6, Pro $16, Business $40 (unlimited seats) |
| Enterprise Security | SOC 2 Type 2, ISO 27001, zero-retention APIs (Enterprise plan) | Enterprise plans available | Enterprise plans available | Enterprise-grade platform | 7-tier role-based access control, SSO (Business plan) |
🛠️ Technical Deep Dive
- Notion AI's knowledge hub utilizes Retrieval-Augmented Generation (RAG) to provide accurate answers by first retrieving relevant information from internal Notion pages and then generating responses based solely on that retrieved context.
- Users of Notion AI can select from various large language models (LLMs), including GPT-5.2, Claude Opus 4.5, or Gemini 3, with an 'Auto' option available to select the most suitable model for a given task.
- The architecture has evolved to support autonomous AI Agents, introduced with Notion 3.0, which can execute multi-step workflows for up to 20 minutes across numerous pages simultaneously.
- Notion AI integrates with external applications such as Slack, Google Drive, GitHub, and Jira, allowing its enterprise search capabilities to pull context and information from across a company's connected tool stack.
- For enterprise plans, Notion AI employs zero-retention APIs, ensuring that customer data is deleted immediately after processing and is not used to train AI models by default, addressing critical security and privacy concerns.
🔮 Future ImplicationsAI analysis grounded in cited sources
AI agents will increasingly handle complex, multi-step workflows autonomously across enterprise tool stacks.
Notion's recent updates (Notion 3.0, 3.2, 3.3) and competitor offerings demonstrate a clear shift towards AI agents that execute tasks and integrate across multiple platforms, moving beyond simple Q&A or content generation.
Organizations will prioritize AI solutions that offer deep integration into existing workflows and provide robust knowledge management capabilities.
The challenges faced by companies in integrating AI across disconnected systems highlight the demand for AI embedded directly into all-in-one workspaces to reduce context-switching and improve productivity.
The strategic value of internal data will increase significantly as AI systems rely on well-structured, accurate organizational knowledge for effective operation.
Osaka Gas's focus on meticulous data organization and the RAG architecture of Notion AI underscore that the quality and structure of internal data directly impact the accuracy and utility of AI outputs, making data governance a critical success factor.
⏳ Timeline
2013-08
Notion Labs Inc. founded
2016-08
Notion 1.0 released
2018-03
Notion 2.0 released
2023-02
Notion AI chatbot service released
2025-09
Notion 3.0 launched, introducing autonomous AI Agents
2026-01
Notion 3.2 released, bringing full Agent functionality to mobile and new model selection
📎 Sources (19)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
📰
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: ITmedia AI+ (日本) ↗