I Gave PodGlue My Own Run of Show

By Junaid Ahmed

---

For 750 episodes I've built my run of show the same way. Hook, promise, introduce the guest. Then numbered segments with times, the questions inside each one, and, this is the part that matters, the exact sentence I say to get from one segment to the next.

PodGlue was generating something that looked close. It wasn't.

Three documents, not one

PodGlue gave me a run of show. It also gave me a list of questions. It also gave me an intro script.

Three separate things. And the questions weren't attached to the segments they belonged to, they were a flat list with timestamps, sitting beside a timeline with its own timestamps.

Which means while I'm recording, I'm the one matching them up. Looking at segment three, scrolling to find which questions have a timestamp near segment three, holding the answer in my head while the guest is mid-sentence.

That's not a run of show. That's raw material for one.

The line I never noticed was missing

Here's the thing that actually bothered me when I looked closely.

In my own run of show, every segment ends with a sentence in quotes. "Okay, so intuition isn't magic, it's signal. Let's talk about why people can't hear the signal."

That sentence is the whole reason the episode doesn't stall. Without it, you get the pause. The guest finishes, you both wait, one of you says "so… yeah." Every podcaster knows that pause. It's the thing you cut in post.

PodGlue was generating those transitions. It had been the whole time. It just never showed them to me anywhere.

What it generates now

I handed PodGlue an actual run of show, the one I used for an episode on intuition, and rebuilt the generator to produce that shape.

So now it writes the opening as three timed blocks. Hook, promise, introduce the guest. Each one is the words I say, not a description of what to say. The guest intro pulls from their real bio, and it's specifically not allowed to invent a credential, if it doesn't know how many clients someone has worked with, it leaves it out rather than guessing a number I'd then say out loud on tape.

Each segment has its time range, its questions, and the transition out. Questions are labeled by what they're for. There's a specific type I care about, the clarifier. That's the one that stops a vague answer. "How do you distinguish intuition from fear or wishful thinking?" It's marked differently because it's load-bearing, not optional.

There's a live exercise segment. A scenarios segment where you apply the idea to three real situations. Numbered homework at the end that someone can actually do this week. And a rapid-fire bank held in reserve for when you're running short, because ending eight minutes early with nothing left to ask is its own kind of bad.

Then I put it on a teleprompter

Same script, full screen, scrolling. Speed control, font size, mirror mode if you're shooting through glass.

One thing I got wrong on the first pass and had to fix: everything looked the same.

A run of show mixes two completely different kinds of text. There's what you say, the hook, the transitions, the sign-off. And what you glance at, the talking points, the things to steer the guest toward.

At 44 pixels on a prompter, those look identical. Which means sooner or later you read your own notes out loud. "Talking points to draw out, chronic fatigue state, energy drain from indecision." On tape.

So they're rendered differently now. What you speak is full size and white. What you glance at is smaller, dimmer, set back behind a line. You can tell at a distance which is which.

You can also send the whole thing out as a text file or an RTF, if you'd rather use the prompter app you already own.

What this actually changed

I ran it on a real episode this week. It works.

The part I didn't expect: I stopped scrolling. The whole episode is one document that reads top to bottom, in the order it happens. When a segment ends, the next thing on screen is the sentence that gets me out of it.

---

The run of show isn't a document you read before recording. It's the thing you record from. Those are different objects, and PodGlue was building the first one while I needed the second.

PodGlue generates the prep so the conversation is the only thing you have to think about.