이미지: METAL LAB 생성
Summary
- On August 18, 2026, OpenAI released three tutorial videos in succession demonstrating how to use ChatGPT Work
- The videos show a sales team using it to prepare for and follow up on client meetings, a marketing team using it to draft blog posts, and a strategy team using it to build strategy decks for leadership reporting
- All three videos link to the ChatGPT Work introduction page (chatgpt.com/work) and build on the recently announced Codex integration and computer history features
- 공개일
- 2026년 8월 18일, 3편 순차 게시
- 영상 1
- 영업팀 Alex — 고객 미팅 브리핑, 후속 노트·이메일·CRM 업데이트 초안
- 영상 2
- 마케팅팀 Sahil — 제품 문서·리서치·회의록을 모아 블로그 초안 작성
- 영상 3
- 전략·사업운영팀 Arvind — 시장 리서치·고객 데이터 종합해 전략 덱 작성
- 공통 진입점
- chatgpt.com/work
- 관련 선행 발표
- 8월 11일 챗GPT 워크-코덱스 동기화, 8월 13일 컴퓨터 히스토리 기능
Five minutes before a meeting, ChatGPT Work fills the gap
Before a sales rep calls a client, they have to dig through account history, and after the call, they have to redo the meeting notes and follow-up emails — a laborious process. In a tutorial video OpenAI posted on August 18, Alex from the sales team hands this process off to ChatGPT Work. Before the call, it pulls together account context to create a meeting brief; after the call, ChatGPT Work drafts the meeting notes, follow-up email, and even the CRM update, leaving Alex only to review them.
The same day, OpenAI posted two more videos in a similar format. Released three minutes apart, all three videos show, through the screens of actual employees, how ChatGPT Work is used in real work.
Background: what ChatGPT Work has been doing
ChatGPT Work is a lineup of ChatGPT products designed for business use. As covered in ChatGPT Work and Codex now automatically sync work from other agents, OpenAI added a feature to the ChatGPT desktop app on August 11 that syncs projects, conversations, skills, and plugins between ChatGPT Work and the coding agent Codex. Two days later, on August 13, it added a "computer history" feature that remembers recent work context to pick up where you left off and suggests skills and scheduled tasks based on repeated patterns. These three videos are essentially a follow-up showing how real work scenarios play out on top of those features.
Three demos, one recurring pattern
The three videos cover different industries but follow the same structure: gather scattered materials, produce output that fits a set format, and have a person review and approve it.
| Tutorial | Person/Team | Input materials | Output |
|---|---|---|---|
| Client meeting prep | Alex, Sales | Account history | Meeting brief, post-call notes/email/CRM update draft |
| Blog draft writing | Sahil, Marketing | Product docs, research, meeting notes, messaging | Launch blog draft (following an existing template) |
| Strategy deck creation | Arvind, Strategy & Business Operations | Market research, customer data, past strategy materials | Strategy deck for leadership reporting |
In the marketing demo, Sahil has ChatGPT Work follow a set blog template and keeps the draft updated as launch details change. In the strategy demo, Arvind personally reviews and revises the storyline and supporting data and figures that ChatGPT Work proposes before approving the final version. All three videos share the common trait of a human giving the output a final check rather than using it as-is.
How to actually try it
All three videos link to chatgpt.com/work. Based on information OpenAI has previously disclosed, ChatGPT Work currently integrates with Codex in the ChatGPT desktop app, letting users import projects, conversations, skills, and plugins, with an option in settings to turn on automatic updates. These videos serve as concrete examples of what kind of output results when that functionality is applied to three specific tasks: preparing for sales calls, drafting blog posts, and building strategy decks. However, the videos do not specify which subscription tiers can use this feature or how availability differs by region.
Editor's view
It's no accident that OpenAI posted three work scenarios side by side on the same day. Choosing three different departments — sales, marketing, and strategy — and attaching a named, real employee to each is a positioning move meant to brand ChatGPT Work not as an "all-purpose chatbot" but as a "department-specific work tool." It's the same playbook competing products like Microsoft Copilot and Salesforce Einstein have used to court enterprise customers with department-specific features. The fact that OpenAI followed up its August 11 Codex sync and August 13 computer history releases with hands-on usage videos right away suggests that OpenAI considers showing how to actually apply features to real work just as important as the pace of stacking new features.
Put into practice, the real value of these department-specific demos lies not in "automatic generation" itself but in the "design that leaves a review step in place." All three videos include a scene where a person double-checks the final output — the CRM update, the blog draft, the figures in the strategy deck — rather than sending it out untouched. It's a sign that OpenAI is well aware that the most common failure in early-stage AI agent adoption is "output sent out without review." If domestic companies want to reference this workflow, the right move is to first copy the design that keeps the final approval step with a human. Expanding the scope of automation is a matter for later.
More department-specific demo videos like these are likely to follow in the coming weeks. Departments such as customer support, HR, and finance have yet to be covered, and OpenAI appears set to keep building up ChatGPT Work use cases ahead of Dev Day season.
