Generate editable PowerPoint files locally from documents or topics
Pass a PDF or document file to an AI, and it produces a presentation that you can freely edit in PowerPoint. This tool uses PowerPoint's native object model to generate shapes, charts, and tables, rather than simple text boxes. The output is an editable .pptx file, not an image, allowing you to fill in new content while preserving existing templates or adding animations and narration.
Using PowerPoint's native objects
Most AI-generated slides consist of text boxes or images, making them difficult to modify, but this tool leverages PowerPoint's native object model directly. Adjustment handles work on shapes and connectors, and charts and tables are created as data-based native objects, enabling the 'Edit Data' feature. It also supports slide master and layout inheritance structures, so you can fill in content while maintaining the design of existing templates that include corporate logos or fonts. SmartArt is intentionally excluded, but a mapping guide is provided to show how close the tool is to PowerPoint's native features, transparently disclosing the current scope of support.
A workflow that runs on your computer
This tool does not rely on cloud platforms; it is a workflow that runs inside an AI tool with agent capabilities. After installing Python and an AI tool, users simply place source materials in the project folder and issue instructions in the AI chat window. Data is processed locally, except for AI model calls and the selected image search service, ensuring security and privacy. It is not tied to a specific platform or model, allowing the use of various AI models such as Claude, GPT, Gemini, and Kimi, with costs limited to AI model usage fees without additional subscription charges.
Installation and execution
After installing Python 3.10 or higher, install dependencies by running the pip install -r requirements.txt command in the project folder. Windows users are provided with a separate guide for additional steps such as PATH configuration. You can choose any agent tool that supports file reading/writing and command execution, such as VS Code, Cursor, or Claude Code CLI. Open the project folder in your AI tool, place PDF or DOCX files in the projects/ directory, and request "Make a PPT from this file." The AI then automatically performs content analysis, visual design, SVG generation, and PPTX export after reviewing the design specifications.
Model selection and quality
This tool handles the workflow, but the quality of the output depends heavily on the capabilities of the AI model used. Using a cheaper model may result in lower completion quality, in which case you should consider upgrading the model first. This tool is not a magic solution that perfectly automates all tasks, but a tool that replaces tedious repetitive work, so users must perform the final finishing touches themselves.
Things to check before use
This tool is released under the MIT license, but PyMuPDF, used for PDF conversion, follows the AGPL-3.0 license, so caution is required when redistributing. SmartArt is not supported, which is an intentional design choice. You should recognize that the output is not a perfect deck completed in one go, but an editable file that users can continue to refine. The quality of the output varies depending on the AI model's context window size and support for image generation features, so it is important to choose a model suitable for your environment.
