We are proud to launch Podhoc, an AI-powered platform that turns written content — articles, PDFs, documents, YouTube videos — into studio-quality podcast episodes in minutes.

Why Podhoc

The idea came from a simple observation: there is too much great written content and too little time to read it. Podcasts solve the time problem, but producing one has always required recording, editing, and hosting. Podhoc removes all of that friction.

How It Works

1. Add a source — paste a URL or upload a document
2. Generate a script — AI reads the content and writes a conversational podcast script
3. Produce the audio — text-to-speech voices bring the script to life
4. Listen or share — stream directly or download the MP3

The entire pipeline runs on AWS: Step Functions orchestrate Lambda functions that handle transcription, script generation (via Google Gemini, OpenAI, and other providers), and audio synthesis.

The Stack

Podhoc is the most ambitious project we have shipped internally:

• Frontend — React 19, TypeScript, Vite, Tailwind CSS
• Auth — AWS Cognito with social login (Google, Apple)
• Backend — Python Lambda functions behind API Gateway
• Orchestration — Step Functions for the multi-stage podcast pipeline
• Storage — S3 for transcripts, scripts, and audio; RDS PostgreSQL for metadata
• Payments — Stripe for credit-based billing

Open for Early Access

Podhoc is live at podhoc.com . Sign up, add a source, and have your first podcast episode in minutes.

Read more on the project page or visit davidpelayo.com for the engineering deep-dive.