Presenton
github.com/presenton/presenton- Category
- AI Tools
- Rank
- No. 126Tools index
- Type
- APP
- Builder
- presenton
- GitHub
- 10.2k stars
- Latest release
- v0.9.10-beta
- Date
About
An open-source AI presentation generator and API — a self-hostable alternative to Gamma and Beautiful.ai.
What it does
Presenton turns prompts, uploaded documents, or existing PowerPoint designs into slide decks. You choose a built-in or custom template, connect text and image providers, edit slides through a drag interface, then export editable PowerPoint or PDF files. It runs as a desktop app, web deployment, or team API.
Why it's ranked here
Presenton covers more of the presentation workflow than generation alone. It accepts source documents and custom designs, supports manual editing, and produces editable PowerPoint files. Broad model-provider support, local model options, an API, and Model Context Protocol access make it useful beyond a single desktop session.
What's good
Provider choice is unusually broad: hosted services, OpenAI-compatible endpoints, Ollama, and LM Studio are supported. Text and image providers can be mixed. Teams can deploy private workspaces with account administration, while designers can build templates with HTML and Tailwind CSS or convert existing PowerPoint designs.
Tradeoffs
Operating it yourself means configuring models, image sources, keys, storage, and deployment. Some providers add their own complexity: Amazon Bedrock requires matching regions, accepted model identifiers or inference profiles, and suitable permissions. Desktop development also needs Node.js, Python, and additional package tooling. Its export targets are PowerPoint and PDF.
How to use it well
Use Presenton when a team repeatedly drafts decks but still needs human editing, house templates, provider choice, or private deployment. Start from a prompt or source document, generate structure and visuals, then finish details in the drag editor or PowerPoint. It does not replace a general document editor or broader design suite.
Technical notes+
The root package.json defines an ES module package with Node-based template conversion and metadata tests, plus Sharp and LlamaIndex LiteParse dependencies. start.js supervises the FastAPI and Next.js services, persists user configuration with a backup and atomic rename, prepares application data directories, and checks the presentation export runtime. electron/build.js packages the desktop application and contains macOS signing, notarization, App Store, and provisioning logic. scripts/convert-template.mjs validates template JSON, rewrites editable image placeholders, copies referenced assets, blocks paths escaping approved roots, and removes stale static assets. test_server.py is only a development stub for Electron version checks.
Observed
- License
- Apache 2.0
- Implementation languages
- JavaScript and Python are both present in the supplied repository files.
- Interfaces
- Web application, Electron desktop application, presentation generation API, and built-in Model Context Protocol server.
- Packaging
- Docker image plus desktop packages for macOS, Windows, and Linux.
- Desktop platforms
- macOS on Apple Silicon or Intel, Windows x64, and Linux x64.
- Export formats
- Editable PowerPoint PPTX and PDF.
- Provider surface
- Supports hosted providers, OpenAI-compatible endpoints, Ollama, and LM Studio.
- Root test surface
- The root npm test script runs template converter tests and a package metadata test.
Read from README.md, package.json, docs/amazon-bedrock.md, docs/multi-docker-build.md, docs/macos/dev/README.md, docs/macos/dev/direct-distribution.md, docs/macos/dev/mac-app-store-setup.md, docs/macos/dev/signed-dmg-build-steps.md, start.js, test_server.py, electron/build.js, scripts/convert-template.mjs, scripts/rebuild_notice_all.py.
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.