presenton/presenton

Presenton – Open‑Source AI Presentation Generator

What it is – Presenton is a self‑hosted web/desktop application that lets you create PowerPoint‑style decks using large language models (LLMs) and image generators. You can feed it a text prompt, an uploaded document, or an existing PPTX template, and it will produce a fully editable .pptx (and PDF) deck.

Key capabilities

• Model‑agnostic: works with any OpenAI‑compatible provider (OpenAI, Anthropic, Gemini, Azure, Bedrock, Ollama, LM Studio, etc.) – you supply your own API key or run a local model.
• Template‑driven: built‑in business, pitch‑deck, and creative templates; you can also create custom templates with HTML/Tailwind CSS.
• Drag‑and‑drop editor: after generation you can fine‑tune slides in a visual editor before export.
• Export: saves as fully editable PowerPoint (.pptx) and PDF, preserving layouts, charts, icons, and images.
• Deploy options:
  • Docker (one‑command container, GPU support available)
  • Electron desktop app for Windows, macOS, Linux (offline use)
  • Cloud SaaS (hosted version) and Enterprise Helm charts for Kubernetes with SSO/SCIM.
• API: a Model Context Protocol (MCP) server lets you call the generation engine programmatically.
• Privacy‑first: all processing can run locally; no data is forced to a third‑party SaaS.

Getting started

1. Docker – run the container and open http://localhost:5001:

   docker run -it --name presenton -p 5001:80 \
     -v "$(pwd)/app_data:/app_data" ghcr.io/presenton/presenton:latest
   

2. Desktop – download the platform‑specific installer (.dmg, .exe, .deb) from the website, then launch the Electron app. The first‑run wizard lets you set environment variables (LLM choice, API keys, etc.).
3. Configuration – set variables like LLM=openai, OPENAI_API_KEY=…, OPENAI_MODEL=gpt-4.1 in a .env file or via the UI. The same variables are read by both Docker and the desktop backend.

Typical workflow

• Choose a template (e.g., Momentum for business reports).
• Provide a prompt such as “Create a 10‑slide deck summarizing Q2 sales performance” or upload a PDF/Word file.
• Select the LLM and image provider (DALL‑E 3, Gemini Flash, Pexels, etc.).
• Let Presenton generate the slides, then edit any content or layout with the drag‑and‑drop editor.
• Export the final deck as .pptx or PDF, or call the /generate endpoint of the MCP API for automation.

Why it matters – It gives teams full control over the AI generation pipeline (no SaaS lock‑in, no hidden fees) while handling the whole presentation lifecycle—from content creation to polished, editable slides. This is especially useful for organizations that need to keep proprietary data on‑premise or want to experiment with different LLM back‑ends.

License – Apache 2.0 (permissive, commercial‑friendly).

---

Quick links

• Quickstart / download: https://presenton.ai/download
• Docs & API reference: https://docs.presenton.ai/
• Docker image: ghcr.io/presenton/presenton
• Community: Discord, YouTube, and public templates.