如果你大量工作是基于Codex，对外分享就变得很简单。 只需让 Codex 整理你的所有对话，从中整理项目和经验。 然后输出为飞书文档和PPT大纲。 有大纲后，…

Codex is an AI collaboration platform by Anthropic that supports long, multi-turn conversations and project management. Its core capabilities include organizing conversation history, extracting key information, and generating structured outputs.

An X user (ID: Xiangyang Qiaomu) shared a workflow: if a user's substantial work is based on Codex, they can use the tool to organize all conversations, extract projects and experiences, and output them as Feishu documents and PPT outlines.

The user describes a three-step process: first, have Codex organize all conversation records; second, extract project key points and lessons learned from the conversations; finally, output in Feishu document format or PPT outline.

For PPT creation, users can employ a custom 'PPT Skill' or directly call Codex's built-in image generation capabilities to transform the outline into a complete presentation.

If this workflow is reliable, it could significantly reduce the cost of converting AI conversations into formal documents, making daily work results easier to share and reuse.

For users whose primary work format is conversation (such as researchers and product managers), Codex's capability could help quickly consolidate experiences into distributable knowledge assets.

This sharing is purely personal experience, lacking third-party validation or official endorsement. Output quality, format compatibility, conversation context completeness, and other key factors have not been evaluated.

The user did not specify thresholds for conversation volume, privacy handling (e.g., whether sensitive data is output), or permission management in cross-team collaboration.

This case suggests AI tools are extending from mere task execution to knowledge management. Conversation history is no longer just a record but a raw material that can be reprocessed.

In the future, similar workflows may become standard for AI-native offices, but current validation from more user practice and product iteration is needed to confirm their universality and reliability.