Did you know that Your presentation and Your AI knowledge source can be the same file.

A developer built MDDeck, a free browser-based tool that keeps Markdown as the single source of truth for both slide presentations and AI knowledge sources, eliminating the need to unpack PPTX files to recover content for agents. The tool renders .md files as slides and adds companion skills that let agents generate MDDeck-format presentations or convert existing PPTX decks into Markdown. All parsing and rendering happens locally in the browser, and no registration is required.

Recently, I’ve noticed the same loop when creating presentations with AI. Usually, we are asking Claude or CoPilot or ChatGPT to create a PowerPoint presentation, spend tokens generating the PPTX, then spend more tokens adjusting fonts, content and layout. Later, the same presentation in "pptx" format is used as a part of knowledge source for an AI Agent or Workflow. The Agent needs to extract the content from the PPTX presentation before using it. That means more processing and, depending on the workflow, more tokens. If the extracted content isn’t saved for reuse, the same work can happen again and again "Reading - Transforming - Spending tokens". We had the information. Turned it into slides. Then asked AI to get the information back out. If we already had the information before we turned it into slides. Why make it harder to use afterward? That’s why, for personal usage, I built https://mddeck.app/ https://mddeck.app/ The idea is simple "Keep Markdown as the source of truth." It lets me keep Markdown as the main source for a presentation. I can render the .md file as slides for a meeting, then use that same file as a text-based knowledge source for an AI agent. When something changes, I update the source. I’ve also added two companion skills to help agents to create presentations in MDDeck’s format or convert existing PPTX decks into Markdown. It has the limitations you’d expect from Markdown. You won’t get complete freedom over slide layouts. Images need accessible URLs or embedded data, and complex charts often work better as separately prepared assets. Converting an existing deck won’t preserve every visual detail, either. And if important knowledge lives inside an image, you still need a text description or the underlying data. Your local files and pasted content stay in your browser. MDDeck parses and renders them locally without uploading their contents elsewhere . Remote images and Markdown links still make requests to their hosting servers. Agents still spend tokens reading Markdown. What MDDeck avoids is the extra work and extra tokens for unpacking a presentation just to get back to its content. Everything is free. No registration is needed.