Description
Podverse is an open-source GitHub project that shows how to add AI features to podcasts. Its own README describes it as "meant as a demonstration of a full-stack webapp that leverages AI capabilities," built by developer Matt Welsh (mdwelsh). The repository is released under the Apache 2.0 license and has been public since February 2024. It is not a hosted product: the project's earlier live demo deployment on Vercel is no longer available, so the code itself is what's maintained today.
Read More
Given a podcast's RSS feed URL, Podverse imports its episodes and runs them through Deepgram for automatic transcription, including diarization (identifying which speaker said what). An LLM then generates episode summaries, and a chatbot built with retrieval-augmented generation (RAG) lets a user ask questions and get answers grounded in the transcript. The app also provides full-text search across transcripts, metadata, and summaries.
Podverse is built with Next.js and designed for serverless deployment on Vercel. It uses Supabase as its backend database, Clerk for authentication, Inngest for background jobs such as processing a newly imported podcast, Deepgram for transcription, OpenAI models for its LLM features, and Mailgun for sending email, with Stripe as an optional billing integration for a deployer who wants to charge for access. The repository includes a full Supabase schema dump (supabase-schema.sql) so a deployer can restore the expected database structure directly rather than recreating it by hand.
Because it ships as source code rather than a running service, Podverse is aimed at developers who want a working reference for building AI-powered audio tools, or who want to self-host their own podcast-AI app. Setting it up requires creating accounts and API keys for several third-party services (Supabase, Clerk, Deepgram, OpenAI, Mailgun, and optionally Stripe) and deploying it yourself; there's no public instance to sign up for. Development activity is light: the repository has a small following (single-digit GitHub stars) and infrequent updates, though it remains unarchived on GitHub.
- CategoryAI Podcast Tools
- Pricing ModelOpen source
- DeploymentVercel
- PlatformsWeb
Podverse's Core Features
RSS feed import to add any podcast's episodes to the app
Automatic transcription and speaker diarization via Deepgram
AI-generated episode summaries
RAG-powered chatbot that answers questions grounded in the episode transcript
Full-text search across transcripts, metadata, and summaries
Serverless Next.js architecture deployable on Vercel
Optional Stripe billing integration for a self-hosted deployment
Open-source codebase (Apache 2.0) for developers to fork and extend
Getting Started with Podverse
Clone the repo: Clone the mdwelsh/podverse repository and run yarn to install dependencies
Provision services: Create accounts for Supabase, Clerk, Deepgram, and OpenAI, and add their API keys as environment variables
Deploy: Push the project to Vercel, adding the Supabase and Inngest integrations to your Vercel project
Import a podcast: Add an RSS feed URL to bring in its episodes for transcription and summarisation
Ask the chatbot: Use the built-in RAG chatbot to ask questions about an episode's content
Podverse's Use Cases
- Podcast transcript search
- Episode Q&A chatbot
- AI reference implementation
- Self-hosted podcast AI
