Generate Podcasts With AI

Jan 20, 2026

I keep up with the AI landscape via AI-generated podcast episodes using the following workflow

1. Gather Topics

Subscribe to TL;DR, take interest in important topics.

Anything you're interested in. If you're curious my method: mostly the TL;DR mailing list, and they have other topics than AI. Eventually I'll automate topic-based content-generation in /tts.

Sample Topics

Lately I'm deep-diving agentic workflows & Claude Code subtopics. I'm exploring the Lang family (LangChain, LangGraph, LangSmith); N8N; and various Claude Code topics like Ralph, Ralph, Skills, Beads, etc.

2. Generate Text

Ask Deep Research to for the scoop

This part's important: use Deep Research, rather than basic chat. Gemini, ChatGPT, Claude, Perplexity - they all have one.

Which Deep Research is best?

If you don't have a subscription, use Gemini, it's free. I personally favor ChatGPT's, but my opinion changes all the time. They're all good, and always leap-frogging.

Where is Deep Research?

Each AI tool has a "Deep Research" or "Research" tool. In the chat area, click the + icon, or the Tools section or such.

Sample Prompt

Replace [TOPIC] with whatever you're learning: "transformer attention mechanisms", "RLHF in language models", etc.

I want to deeply understand [TOPIC]. Write a comprehensive educational guide covering:
- Core concepts and fundamentals
- Historical context and why this matters
- How it works technically (at an intermediate level)
- Key papers, breakthroughs, or milestones
- Current state-of-the-art and open problems
- Practical applications and real-world examples

Write for someone with a technical background who is new to this specific topic.
Be thorough but avoid unnecessary jargon. Aim for 3000-5000 words.
How to Export
  • Claude: export as Markdown. Best format - this is what my TTS tool expects.
  • Gemini: Export to Docs → File → Download as Markdown.
  • ChatGPT: copy the text and paste it directly (my tool takes pasted text too).

3. Convert to Audio

Pipe its output to my Text-to-Speech (TTS) tool as an episode
  1. Go to /tts → Sign Up → create a podcast
  2. Paste the markdown (or upload a .md / .txt) as an episode
  3. Copy the RSS feed into a podcatcher which supports custom feeds
Other TTS Options

If you don't want to use my tool, I recommend ElevenLabs GenFM. ElevenLabs has near-perfect voice realism and is the right pick if you want full-book audiobook narration from EPUB/PDF — a feature my tool no longer supports. I'd personally stop your search there for audiobooks; I went down the Speechify vs Natural Reader vs ... rabbithole, and Eleven wins all day every day. I compare models here (my tool uses Kokoro by default; Qwen3-TTS for voice-cloning).

Want to Use My Voice?

Pay to unlock voices. Then in the voices list you'll find Tyler! Generation costs more credits, since Qwen is more compute-intensive than Kokoro. But the output quality is better, and hey, it's me!