If, you like me, are tired of listening to Alastair Campbell and Rory Stewart agree agreeably on The Rest is Politics, then I might have something for you… 🥁🥁🥁
It’s an 11-part podcast series on Web Development – planned, researched, scripted and read aloud for your entertainment by ROBOTS!!!
Huh?
Well, not robots robots; I’ll explain later.
Here’s a 30-second snippet from the second episode of the podcast:
The full series is available right now on the Internet Archive:
Under the Hood: An AI-Generated Podcast Series on Modern Web Development Concepts
The Internet Archive, by the way, is a fascinating thing in its own right – a nonprofit organisation that archives the internet (you may have heard of its Wayback Machine) and currently has more than 212 petabytes (1 petabyte = 1,000,000 gigabytes) of data stored in its data centres.
I’ve listened to this podcast series in its entirety – the full 6 hours, 47 minutes and 29 seconds of material – and I found it really informative and engaging.
There are some issues, don’t get me wrong. Our virtual podcast hosts have no sense of humour whatsoever, overuse cheesy phrases like “there’s so much to unpack here” and “‘a-ha!’ moments”, mishandle acronyms (either by not introducing them, or by mispronouncing them), speak too fast at times (episode #7 on security is particularly bad for this), repeat themselves (see episode #9 on testing, where our hosts seem perversely determined to drill into you the difference between unit, integration and end-to-end testing) and on one occasion, get stuck in a loop (see episode #10 on DevOps, between 25:30 and 27:15).
BUT, ultimately, these feel like mere quibbles. Because we are talking about a technology that can generate entertaining, factually correct, hour-long podcasts about any subject you can possibly imagine. Just stop and think about that for a minute – it’s kind-of remarkable, isn’t it? There have been a handful of advances in AI over the past couple of years that have really wowed me, and this is definitely one of them.
Not robots robots. So how were these podcasts made?
Step 1 – I started a new conversation with Gemini 2.5 Pro and said this:
I want to generate a long and detailed podcast going through a number of different areas of modern web development, focusing on server-side development but touching on the frontend and integration. It should focus on developing a deep understanding of how things work under the hood, with high level concepts (e.g., HTTP) preferred over diving deep into specific frameworks. Come up with a detailed tutorial regimen, broken down into numbered topics and subtopics, in a logical ordering to take somebody all the way to expert level.
Gemini returned with a detailed, 11-module plan, covering the following topics:
- The Foundation – Anatomy of a Web Request
- The Server-Side Entry Point – Receiving and Understanding Requests
- Processing and State – The Core Logic
- Data Persistence – Storing and Retrieving Information
- Application Architecture & Design Patterns
- Communicating with the Frontend & Other Services
- Security In-Depth
- Performance & Scalability
- Testing Strategies for Server-Side Applications
- Deployment, Operations & Observability (DevOps Mindset)
- Advanced Topics & Future Trends
Step 2 – I started a separate conversation with Gemini 2.5 Pro Deep Research for each module, pasting across the full podcast tutorial regimen and replacing X with the module of interest in the instruction below:
Below is my hypothetical podcast tutorial regimen. I want you to generate a detailed report providing educational information about module X. My plan is to turn your report into its own podcast using AI, so the report should be a rich learning resource in itself, as it’ll provide the source content of the podcast. Focus on module X only.
Gemini returned with detailed educational reports, each one citing between 70 and 261 online sources and ranging in length from 5,051 to 11,529 words.
Step 3 – I exported all of the reports to Google Docs.
Step 4 – I created a separate notebook in Google NotebookLM for each module, importing the relevant reports from Google Docs.
Step 5 – I generated an Audio Overview for each notebook. No custom instructions given.
That’s it! Minimal human involvement, all told. It’s just Deep Research -> NotebookLM Audio Overview. Please note that Gemini Deep Research is only available to Gemini Advanced users; you can get Gemini Advanced via a Google One subscription for £18.99 a month, with the first month free. Or if you don’t want to shell out, OpenAI offer their own Deep Research tool that would probably be similarly effective here, with five queries a month for free.
Conclusion
Since creating this Web Development podcast series, generating and listening to custom podcasts has become a part of my normal routine. I’ve done it for topics including computer networking, UK politics, a comparison of the Nordic nations, hacking, the future of AI, Python internals, and even a feasibility analysis for a business idea I had. If you are like me and listen to podcasts for educational purposes, I highly recommend you give it a go – it’s so easy to do, and it allows you to learn about stuff you’re specifically interested in, rather than having to fish from the limited pool of what’s already available.