How This Show Is Made

Season 2 of The Cynical Developer sounds like me, but I’m not the one at the microphone. That’s an odd thing to admit halfway down a page rather than shout about, so here’s the whole picture in plain terms.

The voice

Every Season 2 episode is voiced using an AI model of my own voice.

The model was trained on recordings of me, and nobody else. It isn’t an impression of another presenter, it isn’t a stock narrator, and it isn’t pretending to be someone who doesn’t exist. It’s my voice, generated rather than recorded.

You’ll hear this said out loud at the end of every Season 2 episode, and it’s written into the episode notes and the audio file’s metadata too. If you’re listening in a podcast app, you shouldn’t have to come to this page to find out.

The scripts

I write the episodes. Every topic, every argument, every opinion you disagree with — mine.

AI helps me tighten them. It cuts the waffle, flags where I’ve made the same point twice, and tells me when a paragraph isn’t landing. It’s a good editor and a poor writer, so that’s the job I give it. Nothing goes out that I haven’t read, changed, and signed off.

Season 1 is a different thing entirely

Everything before Season 2 — all 159 episodes — is real recordings of real conversations with real guests. No AI voices, no synthesis, nothing generated. If you’re listening to an interview on this show, you’re hearing the people who were actually in it.

That distinction matters enough that I’d rather spell it out than let one page cover both.

Why do it this way

Honestly? Because the alternative was not making the show.

Season 1 ran on a format that needed scheduling, guests, travel and a studio. Season 2 is a short weekly monologue, and producing it the traditional way means finding a clear hour with decent acoustics every single week. That’s the thing that killed the release schedule the first time round.

Writing stays hard and stays mine. Recording stopped being the bottleneck. The show ships weekly again as a direct result.

What I won’t do with it

I won’t use the voice model to say things I didn’t write. I won’t put words in anyone else’s mouth, synthesised or otherwise. I won’t run ads or sponsor reads in the AI voice without saying so. And I won’t quietly drop the disclosure once people are used to the sound of it.

If that ever changes, this page changes first.