Select a track that matches your goal — pitch deck, thesis defense, product launch, and more. This shapes the slide structure and terminology.
No Source Material Attached
The model will synthesize the blueprint from general knowledge and your topic alone. Results may lack specificity and data grounding. For best results, paste relevant source material.
Keyboard Shortcuts
⌘ EnterCompile Blueprint
⌘ CCopy Payload (no selection)
⌘ EExport Markdown
1 / 2 / 3Switch output tabs
?Toggle this overlay
EscClose overlay / modal
S
Slide OS
v0.2 AgenticClient-Side
Claude
Generate a presentation blueprint in 30 seconds
Use Case
→
Topic
→
Sources
→
Blueprint
No Blueprint Generated
Configure parameters and compile — or try an example below.
Core topic required
Enter a one-sentence summary of your presentation's main argument in the Core Topic field, then compile again.
Payload too large for a single API call
Split your source material into sections and compile twice — Part A for slides 1-N, Part B for slides N+1-M. Or reduce source length and slide count.
Rate this output:
—
Thanks — feedback recorded
Payload copied → Paste intoClaude.aito generate your slides
Generation History
No generations yet
Your compiled blueprints will appear here. Click any item to reload.
TL;DR: Slide OS is a free, client-side tool that generates structured presentation prompt blueprints optimized for Claude (XML + chain-of-thought) and Gemini (hierarchical long-context). Paste your topic and source material, configure audience and tone, and get a copy-ready prompt in under 5 seconds. Supports 7 presentation tracks. Zero data sent to any server — everything runs in your browser.
Maintained by the Slide OS project · Last updated · Status Active development (v0.2 Agentic)
Frequently Asked Questions
What is Slide OS and how does it work?
Slide OS is a free, browser-based tool that generates structured presentation prompt blueprints for AI models like Claude and Gemini. You configure a presentation track (pitch deck, academic thesis, product launch, etc.), enter your topic, paste source material, and click "Compile Blueprint." The tool instantly produces a structured prompt payload — a detailed instruction set you paste directly into Claude or Gemini to generate your full slide deck. Everything runs client-side in your browser with no data sent to any server. It supports two prompt architectures: Claude-optimized XML with chain-of-thought reasoning, and Gemini-optimized hierarchical markdown with long-context anchoring. Output includes slide-by-slide structure with titles, layouts, speaker notes, visual cues, and transition logic.
What is the difference between Claude and Gemini prompt modes?
Claude and Gemini process information differently, so Slide OS generates distinct prompt formats for each. Claude mode produces instructions wrapped in strict XML schema with explicit thinking blocks that encourage chain-of-thought reasoning before output generation. This plays to Claude's strength in structured instruction following. Gemini mode uses hierarchical markdown with phase-based execution — Context Synthesis, Slide Generation, Quality Audit, and Executive Summary. It leverages Gemini's long-context window by anchoring instructions with metadata tables and structured data blocks. Both modes produce equivalent slide deck structures. The difference is in how the instruction is formatted to maximize each model's accuracy and reduce hallucination risk.
Is my source material sent to any server?
No. Slide OS runs entirely in your browser as a single HTML file with JavaScript. Your topic, source material, and generated prompts never leave your device. There are no API calls, no analytics servers, no tracking pixels, and no cookies. The only data persistence is localStorage on your own machine for generation history and basic usage counters. You can verify this by opening your browser's Network tab — you will see zero outbound requests related to your content. For maximum privacy, you can download the HTML file and run it completely offline.
How do I use the generated prompt with Claude or Gemini?
After clicking Compile Blueprint, click Copy to copy the structured prompt to your clipboard. Then open Claude at claude.ai or Gemini at gemini.google.com in another tab and paste the prompt directly into the chat input. The AI model will follow the structured instructions to generate your complete slide deck including titles, layouts, speaker notes, visual suggestions, and narrative flow. For best results, use the generated prompt with Claude 3.5 Sonnet or Gemini 1.5 Pro. You can also export the prompt as a Markdown file for version control or team sharing.
Can I use generated prompts for commercial presentations?
Yes. The prompts generated by Slide OS are instructional text containing no copyrighted content and imposing no usage restrictions. You own the output produced by the AI model when you use these prompts. However, always verify that the source material you paste into the tool is content you have the right to use. The tool's role is to structure your existing information into an effective presentation format. It does not generate original factual claims or proprietary data.
What makes this different from asking an AI directly?
Asking Claude or Gemini to "make a presentation about X" produces generic, unstructured results. Slide OS generates a detailed engineering prompt that specifies slide count, narrative arc, audience calibration, tone, data density, visual layout directions, speaker scripts, and transition logic — all derived from your actual source material. The prompt includes built-in quality audits, anti-hallucination rules, and self-verification steps. Think of it as the difference between asking someone to write an essay and providing a detailed brief with structure, constraints, and source citations. The output quality difference is substantial.