Content and publishing · Edition No. 33 · 27 Sep 2026

presenton/presenton

Generates editable PowerPoint decks from a prompt or a document, on your own machine, using whichever model provider you point it at.

← Content and publishingRead the whole edition →

10,754 stars · Apache-2.0, read from /blob/main/LICENSE on 2026-09-27; plain, with no added conditions · electron-v0.9.11-beta (2026-09-21), read from /releases/latest and confirmed by ungh · Track this in Scout

Generates editable PowerPoint decks from a prompt or a document, on your own machine, using whichever model provider you point it at.

▶Repo detailsthe review · specs · pros & cons · install

What it is

Presenton runs as a small web application inside a container and turns a text prompt or an uploaded file into a presentation. It also ships desktop builds for macOS, Windows and Linux for people who would rather not touch a command line.

What it is good for. Anyone who makes the same shape of deck over and over — a weekly report, a course module, a client summary — and pays a per-seat subscription to a hosted deck generator for the privilege. The output is a real PPTX file rather than a locked page, so it opens in PowerPoint, Keynote or LibreOffice and can be finished by hand. The same generation path is exposed as a single web address, /api/v1/ppt/presentation/generate, so it can be wired into an existing pipeline instead of being clicked.

Stars10,754
LicenceApache-2.0, read from /blob/main/LICENSE on 2026-09-27; plain, with no added conditions
Latestelectron-v0.9.11-beta (2026-09-21), read from /releases/latest and confirmed by ungh
Good
  • Plain Apache 2.0 with nothing added. In a category full of restrictive terms, this one has none.
  • Genuinely free in its choice of model. OpenAI, Anthropic, Google, Azure, Bedrock, OpenRouter and several others are supported, and so are Ollama and LM Studio, which run a model on your own hardware with no bill and no data leaving the machine.
  • The output is an editable PPTX and the interface is one documented endpoint, so it fits into work you already do rather than replacing it.
Watch for
  • Every release is a beta. The whole published history is 0.9.x-beta and there is no 1.0. There is no stable version to pin to.
  • No resource figures are published anywhere. The README, the documentation site and the hosting page give no memory, disk or graphics-card numbers. If you take the local-model route, the model download is the dominant cost and its size is not documented at all.
  • The setup is a command line, a container, a mounted folder and an API key held as an environment variable. That is a real barrier for someone who does not use a terminal, and the desktop build is the only way around it. Everything lives in the mounted app_data folder, so losing that folder loses every deck.
Similar repositories
Install
mkdir -p app_data
docker run --detach --name presenton \
  --publish 5001:80 \
  --volume "./app_data:/app_data" \
  --env LLM=openai \
  --env OPENAI_API_KEY="YOUR_API_KEY" \
  --env OPENAI_MODEL=gpt-4.1 \
  ghcr.io/presenton/presenton:latest
Screenshots
presenton/presenton: GitHub preview cardpresenton/presenton: Screenshot 1presenton/presenton: Screenshot 2presenton/presenton: Screenshot 3presenton/presenton: Screenshot 4

Get the next edition in your inbox

A dozen repositories, opened and checked. The licence read, the last release dated, and the ones that did not make it named with the reason. It is the half most lists leave out.

No tracking pixels. One click to leave. The archive stays free either way.

We use your address to send the edition and nothing else. Confirm by email, leave in one click. How we handle it.