How to turn meeting recordings into LinkedIn content

Updated 24 September 2026 · 8 minute read

Granola, Fathom, Fireflies, whatever you run, it's been sitting in your meetings and capturing everything for months now.

If you record every call and nothing ever comes of it, there are usually three reasons, and none of them have much to do with writing.

Why nothing comes of your recordings: three foundational mistakes

The first is that you're not connecting those calls to other systems, or giving other systems access to them. So they sit in a silo and either get lost or simply never get used.

The second is that you're probably not embedding them, which makes searching those calls very difficult for anything else. Particularly if it isn't possible to run a vector search over them and therefore understand the semantic relationships across potentially thousands or tens of thousands of hours of recordings.

You can't simply search via SQL in those circumstances. Not having a proper process to embed and codify those calls is a critical thing, and it's the one most people never get to.

The third is not knowing what you want out of those calls in the first place. Thinking about the use cases where they could be handy. Figuring out a system to codify the most important bits for easy and fast retrieval. Without that there's no route to the good material even when you know it's in there.

Those are the foundational mistakes. They all sit well before the point where you'd start thinking about content at all.

Label the calls as they come in

It's also important to label the different calls, and where they're going to be used.

Discovery calls are one set, and they're incredibly useful. Demo calls are very useful. Onboarding calls have their own sort of shape and intelligence to them. Customer delivery calls are also really useful, as are the calls where you go through your proposal. All of these are useful for an independent or fractional executive looking to mine different insights, and they don't give you the same kind of insight as each other, which is the reason to label them rather than treat them as one undifferentiated pile.

The important thing is to make sure you're pulling out only the important insights, because maybe 90% of those calls, on a delivery call let's say, are routine or mundane or relatively repetitive. They're not always going to drive insights.

Maybe 90% of any given call is routine, mundane or relatively repetitive. You never know which call the other 10% turns up in.

But you never know during any given call when something really interesting might crop up. Or you might not have noticed that across three or four different clients, as you're doing these repeat calls, there's a pattern emerging. That's where having a system to mine those patterns and look for the trends is useful, because you're not going to spot it yourself while you're in the middle of the work.

Why a transcript beats writing from scratch

A few reasons.

The first is that you actually get to hear the genuine words and phrasing that your clients, in other words your target prospects, use to describe their own situations. Which is far more powerful than any type of keyword research or chatbot research you could do.

Having them articulate in their own words the challenges they have, their goals and opportunities and so on, gives you a genuinely unique perspective. It gives you genuine alpha in the market in terms of what they care about, and therefore what you should write about, and then how to write about it in a way that's going to resonate with them.

It's also really good for hearing how you describe your own point of view. The way you solve those challenges, how you've seen them before and overcome them, how you'd recommend somebody goes about solving them. These are all proprietary to you, which makes your content stand out and be more authentic than if you were just trying to have an LLM produce it for you, or a ghostwriter or copywriter who isn't you.

You're literally using your own words. And often those transcripts have a rhythm and a structure about them that is just different to when you're typing on a blank page.

Is it a post, a note, or nothing?

If I'm looking at a call manually and trying to decide which of the three it is, I'm asking how substantive the topic is. Could you go deeper on it? Did they go deeper on it?

For example, if someone says they're struggling with SEO, and the response is "oh no, let's revisit that", that's probably not a post in and of itself. However, if you double-click on it and say tell me more about the challenge with SEO, and 15 or 20 minutes later you've got an entire call going into a lot of detail around the challenges of modern SEO, that is a post.

Another thing I'd look for is novelty. If someone mentions an unusual situation, or a situation you've not seen or heard of before, that often makes for an interesting context for a unique post.

Then there's repetition across calls. If you hear the same thing across multiple calls, that's usually cause to say right, this is being repeated, people are dealing with this now, and that's definitely a post.

But in any one call, one mention doesn't make a trend.

One mention doesn't make a trend. That's a fragment, and a fragment is a note for later. Hear it enough times and those notes come together and make the post.

So that's where you've got a fragment, and a fragment is a note for later. If you hear it enough times, those notes come together and make the post. Which means the whole thing depends on filing the fragments somewhere you'll see them again.

What I've changed my mind about

I've been using call transcripts to drive my own content through Inklined, and the thing I've changed my mind about is how valuable my own perspectives on those calls have been.

I originally thought most of the value was going to come from hearing the market feedback and the direct phrasing of clients and prospects. What actually happened was that my calls helped seed my fingerprint. Ideas around taste and beliefs and frameworks, without me having to think ahead on those pieces. That's been really valuable, and it's meant my writing has become more distinctive and more original than it was.

The other thing is the sheer volume of interesting ideas and topics that come out of those calls, which is way more than I'd originally anticipated. It's a richer source of inspiration than I thought it was going to be.

What the tools get wrong

Three things, as far as I can see.

Privacy. They don't take privacy and anonymisation seriously enough. It's potentially too easy for key identifiable details to slip through the cracks and make it into the public domain, which for anyone writing about client work is the thing that actually matters.

Storage. I don't exactly know how they store those calls, but there's a good chance they're making the patterns hard to recognise and pull out. Which means quite a lot of potential insights and topics get lost from call to call. That's the cross-call pattern problem again, and it's the most valuable material there is.

Accumulation. As far as I'm aware they're not using any of it to further seed the knowledge base, or the fingerprint, or their equivalent of it, to improve the overall authenticity and value of the writing that gets produced at the end of the day. So the tenth thing you publish is no more yours than the first was.

Where Inklined fits

Inklined reads the calls you already record, from Granola, Fathom, Fireflies or Attio, and you only need one of them connected.

Every transcript is anonymised on the way in, before anything else happens to it. Names become role descriptors, companies become sector descriptors, and contact details are removed. Only your own speech can become something you publish, and anything that still looks like it names an organisation asks you before it will draft from it.

The rest of it does roughly what's described above, except it doesn't forget between calls. Fragments are held until they turn into patterns, and what you say on calls feeds the profile of how you write.

If you're already recording and nothing's coming of it, start with the strategy session, or have a look at how the recorder connections work.

Common questions

Which calls are worth recording for content?

Most of them, and what matters more is labelling the different calls as they come in. Discovery calls are one set, and they're incredibly useful. Demo calls are very useful. Onboarding calls have their own sort of shape and intelligence to them. Customer delivery calls are also really useful, as are the calls where you go through your proposal. They don't give you the same kind of insight as each other, which is the reason to label them rather than treat them as one undifferentiated pile.

How do I turn a call transcript into a LinkedIn post without breaching client confidentiality?

Strip the identifying detail before the transcript goes anywhere near a draft, and publish only your own speech. That means names, the company, sector detail specific enough to point at one business, and any numbers only one client would have. Anonymising at the end, or catching it in review, is not the same thing, because by then the detail has already been through everything you use to write.

Is it safe to put client call transcripts into an AI tool?

It depends on when the anonymisation happens, which is the question worth asking any tool you are considering. If it anonymises on the way in, before anything else touches the transcript, identifiable detail never reaches the draft in the first place. If it happens at the end, or is left to you to catch, then it is potentially too easy for key identifiable details to slip through the cracks and make it into the public domain.

How much of a call is actually usable?

Maybe 10%, and on a delivery call less than that. The rest is routine or mundane or relatively repetitive, and it isn't always going to drive insights. But you never know during any given call when something really interesting might crop up, which is why recording selectively tends not to work.

Do I need a tool for this, or can I do it by hand?

By hand works at low volume. Read your own half of the transcript, mark the places where you made a case or where a client used a phrase you would never have chosen yourself, and file the fragments somewhere you will see them again. What breaks first is the pattern across calls, because you're not going to spot it yourself while you're in the middle of the work. The second thing that breaks is the filing.