Skip to main content

Get started

Daily Dispatch is a set of parts, each of which runs on its own. This page gets each one running on your machine with sample data. None of the commands here need an account, an API key or a deployed backend.

  • The app (app/) shows the day's edition on a phone, an e-ink tablet, the web and the desktop.
  • The daily-dispatch CLI (packages/daily-dispatch/) writes a two-host podcast script from the day's stories and voices it.
  • The pipeline (pipeline/) captures the links you push from your phone and turns them into the day's digest.
  • The backend (backend/) stores editions, episodes, settings and push devices on AWS. You only need it to sync the app across devices or to get push notifications.

Prerequisites​

  • Git
  • Node.js 24. The CLI alone needs 22.18 or newer.
  • pnpm 10 or newer
  • Python 3.12 or newer, for the pipeline
  • ffmpeg, only when you make podcast audio
  • JDK 17 and an Android SDK, only when you build the Android app

Clone the repository:

git clone https://github.com/blbaker/daily-dispatch.git
cd daily-dispatch

Open the desktop app with sample editions​

The quickest way to see the app is the desktop build. It starts a local mock of the backend that serves three sample editions.

cd app
npm run desktop:install
npm run desktop:demo

This works on macOS (Apple silicon), Linux x64 and Windows x64. Linux needs an X11 or XWayland session and Mesa; Desktop lists the packages.

Run the app in a browser​

cd app
npm install --legacy-peer-deps
cp .env.example .env
npm run web

With EXPO_PUBLIC_API_URL and EXPO_PUBLIC_API_KEY empty, the app opens but shows "API not configured". Fill them in once you have a backend (see Backend). Set EXPO_PUBLIC_EINK=true for the e-ink build, which uses a flat grayscale theme.

npm run android and npm run ios build and start the native app on a device or emulator. They need the Android SDK or Xcode.

Make a podcast script without spending anything​

Build the CLI from the workspace and run it on the bundled example stories:

pnpm install
pnpm --filter @blbaker/daily-dispatch build
node packages/daily-dispatch/dist/cli/bin.js podcast make \
  --stories packages/daily-dispatch/examples/stories.example.json \
  --writer anthropic --voice elevenlabs \
  --voices packages/daily-dispatch/examples/voices.elevenlabs.example.json \
  --out /tmp/episode

Both providers in that command are paid, so the CLI calls neither. It saves the prompt the writer would get to /tmp/episode/prompt.txt and prints the character count and an estimated cost. Swap in --writer ollama and --voice local to make a full episode for free with a local Ollama model and a local TTS server. Podcast providers covers every option.

Check the pipeline​

python3 -m unittest discover -s tests
python3 pipeline/podcast_make.py --script-dir pipeline/examples/link-talk-script \
  --voices pipeline/voices.example.json \
  --out /tmp/episode.mp3 --transcript /tmp/episode.md --dry-run

The dry run parses the example script, writes the transcript and prints the character count. It never calls ElevenLabs. Pipeline explains how to capture real links and write a digest.

Next steps​

  • Pipeline: capture links and write the daily digest.
  • Backend: deploy the AWS stack so the app can sync.
  • Daily episode: make a podcast at the same time every day.
  • Push setup: turn on notifications for a new edition or episode.
  • Desktop: package the desktop app for macOS, Linux or Windows.