This Langflow flow implements an automated Gmail → script → audio → Drive pipeline.

It starts by connecting to Gmail and fetching unread newsletters (or messages matching a label/filter). The workflow extracts the relevant body content, removes boilerplate, and normalizes formatting so the source material is clean and consistent.

A script generation stage uses an LLM to transform the newsletter into a spoken format. The flow can enforce a consistent structure (intro, headlines, key takeaways, closing) and tone (neutral, conversational, executive) while preserving factual accuracy.

Next, the text-to-speech stage sends the script to ElevenLabs and generates an audio file using a selected voice, pacing, and audio settings. The workflow can also generate metadata (title, episode description, source links) for organization and reuse.

Finally, an export stage saves the audio output to a Google Drive folder with consistent naming conventions and optional tagging, enabling easy playback and sharing across devices and teams.

The flow can be extended for higher scale and reliability. Add deduplication and episode tracking so already-processed newsletters are skipped, and implement batching to generate a single daily or weekly compilation episode. You can integrate with podcast hosting platforms, Slack, or email to distribute generated audio automatically, and store transcripts and episode metadata in a database for search. For quality control, add a validation step that checks for hallucinations, broken links, or missing context before audio generation. Advanced setups can support multiple voices, multilingual conversion, and personalization rules per sender or newsletter category.