If Your Expertise Never Gets Published, the Bottleneck Probably Isn't Writing

Here is the part that should matter to you even if you never write a word publicly: most of the knowledge inside an organisation never gets published, and it almost never stops because the person can't write. It stops because of friction. The research takes hours. The recording setup takes effort. The editing takes a night. So the expertise stays in someone's head, and the company, the industry, or the community never gets it.

I've spent the last couple of weeks building something to remove that friction for myself, and this article is the first thing produced with it. I want to walk you through how it works, and more importantly, where I drew the line between a tool that helps me write and a tool that writes for me. That line is the whole story, and it's the question I'd want answered if someone brought a system like this to my department.

What the old process actually cost me

I used to read the news for hours at a time to stay current. I'd research what was happening, follow the social media channels, read what my friends in the industry were doing, and form my own take. Then I'd write it out, or think it out, or scribble notes. Then I'd sit down in front of a microphone or a camera and go through all of it, purely because I enjoyed talking about it and presenting my angle.

I got into blogging because Tom Hollingsworth told me to. He was the one who said he thought people would value my opinion. I had honestly never considered that before he said it.

Why it stopped

Over time, something you enjoy becomes easy to look past. It's either too early in the morning or too late at night. Or nobody's online right now so they won't read it. Or I don't have time to edit it. Or I didn't do my hair, or I'm not wearing the right shirt. There was always an excuse for why I couldn't put down a blog post or take the time to record a podcast.

Notice that none of those excuses are about having nothing to say. They're all logistics. Logistics are automatable.

The daily phone call

So I built a system that calls me on the phone at a set time every day, ready to talk about a handful of topics.

Behind that call, it's monitoring roughly 500 news sources through RSS, that's "Really Simple Syndication," a plain machine-readable feed most publications quietly offer so software can pull their new headlines automatically, without a person visiting the site. On top of that, numerous trade magazines and eight different social media channels feed in. The system synthesises all of that down into a short list of top-line story ideas, and I pick one. That single step replaces the hours of reading.

Then it interviews me about my pick. The system asks the questions. It responds to what I say and follows up, the way a real interviewer would rather than just reading from a script. It records my voice, exactly as it's doing right now, and when we're finished it produces a transcript, a written record of everything I actually said out loud. That transcript is then turned into an article written in my voice, the way I would have written it myself.

The line between assisted and generated

This is the part worth scrutinising, because "AI wrote it" and "AI helped me publish it" are very different products with very different risks.

When the draft is finished, I go in, read it, and approve it. I'm approving that it sounds like me, that it talks like me, and that the words in it are genuinely my words from the interview. It does not invent sentences on my behalf. It does not invent storylines for me. It might fix a grammatical error here and there. But it honestly uses my voice and my words to build content that is my content.

The important structural point is that the human approval step is not optional decoration at the end. It's the thing that keeps the output mine. Strip that out and you have a different tool entirely, one I wouldn't want my name on.

Why I'm opening it to other people

I want to share it because I've made it easier for me, and if I can make it easier for me, I can certainly make it easier for you. And if you've never done anything like this, it's a good way to start getting your opinions out there. Instead of having to come up with everything yourself, the system prompts you. It asks you. It interviews you.

It's built around who you are. It understands who you are, what you know, and how to shape the questions that make the most sense for you. It draws out your voice, so the article sounds not only like you, but like the best version of you.

I think it's phenomenal, and I can't wait to share it. To start, I'm opening it up by invitation to a select group of people rather than to everyone at once.

What you can do with this

• Write down the excuse you actually use for not publishing, the real one, about timing or editing or how you look on camera, then work out which part of it a system could take off your hands.
• If you've never blogged, stop trying to originate everything from a blank page and let something prompt and interview you instead.
• Ask, of any AI writing tool put in front of you: does it invent sentences and storylines, or does it only reorganise words the person actually said? Ask to see the transcript alongside the draft.
• Ask who signs off before publication, and whether that approval step can be skipped or bypassed under deadline pressure. If it can, that's your exposure.
• Ask where the source material comes from and how much of it there is, in my case around 500 RSS feeds, trade press and eight social channels, because the quality of the prompt depends entirely on the quality of the inputs.
• Whatever the draft comes back as, read it and approve it yourself before it goes out, so the words stay yours.
• Visit telepost.ai to see what the workflow looks like, and email me if you'd like to be considered for an invitation.

Want this kind of thing weekly, in audio? That's the show.