The short version: One recorded sales or discovery call, run through a transcript tool and then a proper edit, can produce five to seven pieces of content in under two hours. The AI does the drafting. You do the editing, and skipping that step is why so much AI-repurposed content reads flat and gets ignored.
Related reading: high ticket sales low ticket.
The content you already have and never use
Every small business owner I work with has the same problem sitting in their calendar right now. Dozens of recorded calls. Discovery calls, sales calls, client onboarding calls, even the twenty-minute “quick chat” that turned into forty minutes of gold. Real objections. Real language customers use. Real proof that what you sell solves a problem, worked out through someone else’s mouth instead of yours.
And almost none of it gets used. It sits in Zoom’s cloud storage or a Google Drive folder named “Calls 2026” and dies there.
I had a client, a bookkeeper in Manchester called Nadia, who had one hundred and forty recorded discovery calls in her Zoom account going back three years. Three years of small business owners telling her, in their own words, exactly why they were frightened of their own accounts. That’s not a content gap. That’s a content goldmine nobody had opened.
What I do every Friday now
I block ninety minutes on a Friday afternoon and I run the same process on whichever call from that week had the best moment in it. Not the best call overall, the best moment. A sharp objection, a surprising question, a client saying something in a way I never would have written myself.
The steps, in order:
- Record the call with consent, using Fathom, Fireflies or Otter, all of which give you a transcript within minutes of hanging up
- Skim the transcript once and mark the three or four moments with a timestamp, not the whole call
- Drop those marked sections, not the full transcript, into ChatGPT or Claude with a specific prompt: “Turn this into a LinkedIn post in my voice, keep the client’s exact phrase where marked, do not soften it”
- Get a first draft back in under a minute
- Rewrite the first two lines completely by hand, because AI openings are always the weakest part
- Cut anything that sounds like a training manual, usually one sentence in three
- Post it, then repeat for a short video script, a newsletter paragraph and one FAQ answer for the website
That ninety-minute block regularly produces a LinkedIn post, a client email, a short video script for Reels or a talking-head clip, one FAQ answer for the website, and a paragraph for the next newsletter. Five pieces from one conversation that already happened, that you were paying for anyway in the time it took to have it.
The uncomfortable part nobody puts in the “how to repurpose content with AI” articles
Here is the bit that gets left out of most guides on this. The transcript-to-post pipeline works brilliantly for speed and terribly for quality if you publish the first draft. I have read hundreds of LinkedIn posts in the last year that were clearly generated from a call transcript and never touched again afterwards. You can spot them in three seconds. Same rhythm, same three-point structure, same slightly too-neat ending line. They get polite likes from people who owe you a favour and nothing else. The technology was never the hard part. Editing is the hard part, and it always was, long before AI existed. AI just removed the excuse of “I don’t have time to write a first draft,” which means the only thing separating good repurposed content from forgettable repurposed content now is whether you were willing to spend fifteen minutes making it sound like you again.
Nadia’s best-performing post of the year came from a client on a call saying, almost as an aside, “I thought bookkeeping was just typing numbers into boxes until you showed me it was basically insurance against a bad year.” The AI draft turned that into a bland three-paragraph post about “the value of proactive financial management.” I put the client’s actual sentence back as the opening line, cut the AI’s summary underneath it to one line, and it became her highest-engagement post of 2026, with three direct enquiries from it. The AI wrote the scaffolding. The client’s own words, which the AI had quietly smoothed away, were the entire point.
Why this beats “content ideas” tools
Most AI content advice for small business owners still starts with “ask ChatGPT for ten content ideas about your industry.” That produces generic ideas because you gave it a generic prompt. A call transcript is the opposite of generic. It has a real person’s actual words, actual hesitation, actual relief when the penny drops. You’re not asking AI to invent something interesting, you’re asking it to organise something interesting that already happened, which is a much easier and much more honest job for the tool to do well. This is the same instinct behind good brand storytelling generally. Look at how the Away marketing strategy was built almost entirely on real customer travel stories rather than product features, or how Disney’s approach, covered in this Disney marketing strategy breakdown, leans on specific, lived moments rather than abstract brand language. The call transcript is your version of that raw material. Most small businesses are sitting on hours of it and reaching for a generic prompt instead.
What one good call can become
To make this concrete, here’s what came out of a single 38-minute discovery call I ran with a prospective client in January, before I’d even signed them:
- One LinkedIn post built around their exact objection (“we tried an agency before and it was a black box”)
- One 90-second video script answering that same objection on camera
- One paragraph for my newsletter under a section I call “what people say to me”
- One new FAQ answer for my services page, because if one prospect asked it, others are thinking it and not asking
- One cold outreach line I now use with similar prospects, because it tested the phrasing that landed
Five assets, none of them invented, all from a call I had anyway. If you’re building anything interactive on your site to capture that same kind of real objection at scale, the same logic applies to tools like a lead-generating interactive calculator, where the questions people type in are just as valuable a data source as the answer you give them.
The permission and privacy step you cannot skip
Say it on the call. “Do you mind if I record this so I don’t miss anything?” takes four seconds and every single tool I’ve mentioned asks for the same thing anyway. If you plan to quote a client publicly, even anonymised, ask separately afterwards. I keep a simple rule with clients like Nadia: anything used publicly gets the person’s first name changed and their business identifying details removed unless they’ve explicitly said it’s fine to use their name. Two clients have said yes to full attribution because being quoted turned into free exposure for them too. That’s a good outcome for everyone, but it only happens because I asked instead of assuming.
Where this connects to bigger content strategy
None of this replaces having an actual content strategy. It replaces the blank page problem, which is the thing that kills most small business content calendars by week six. Ahrefs built an enormous amount of organic traffic partly by turning real customer questions and real support tickets into content, a pattern covered well in this Ahrefs marketing strategy analysis. The mechanism is the same one I’m describing here, just running through a support inbox instead of a sales call. Wherever real customers are talking to you, in whatever channel, that’s your content source. AI just makes turning it around fast enough that you do it before the moment goes cold.
The mindset shift that matters more than the tools
There’s an old idea in Napoleon Hill’s writing that comes up a lot in my own head when I’m tempted to skip the Friday block: definiteness of purpose, doing the unglamorous repeatable thing on a schedule rather than waiting for motivation. It’s covered in this piece on business lessons from Napoleon Hill. Repurposing calls into content isn’t clever. It’s a habit you either build into your Friday or you don’t, and the businesses that stick with it for three months consistently outperform the ones chasing the perfect AI prompt for a week and then giving up. Howard Schultz built a good chunk of Starbucks’ early brand loyalty on paying close attention to what individual customers said in individual stores, a habit examined in this piece on business lessons from Howard Schultz. The lesson transfers directly. Your customers are already telling you what to write. The call transcript is just the modern version of standing behind the counter and listening.
What I’d tell you to do this week
Pick one call from the last fortnight, ideally one where a client said something you wouldn’t have written yourself. Run the transcript through ChatGPT with the exact prompt above. Rewrite the opening two lines by hand. Post it. Time the whole thing. Most people find it takes under twenty-five minutes once they’ve done it twice, and most people also find the post outperforms whatever they’d have written from scratch that same week, because it started from something true instead of something invented.
Frequently asked questions
Do I need expensive software to repurpose call transcripts with AI?
No. Fathom is free for basic transcription, Otter has a workable free tier, and ChatGPT’s free version handles the drafting step fine for most small business volumes. The whole system can run at no extra cost beyond what most people already pay for video calling.
How much time does this save compared to writing content from scratch?
Most small business owners I’ve worked with go from roughly four hours to produce a week of content down to around ninety minutes, once they build the habit of marking timestamps rather than reading full transcripts.
Is it legal to use client call content in my marketing?
Ask on the call before recording, and ask again separately if you plan to quote anyone publicly, even anonymised. This isn’t just good manners, most call recording tools and platforms require two-party consent to be disclosed anyway, so build the question into your opening line every time.
What’s the biggest mistake people make with this method?
Publishing the AI’s first draft without editing it. The draft gets the structure right and the client’s actual language wrong almost every time, smoothing away the exact phrase that made the moment worth repurposing in the first place. Always put the real quote back in.
Free resource: grab The LinkedIn Post Prompt Pack from the resource library.
Want this done for you? See AI automation for coaches and consultants.
Related reading: AI News This Week for Business, 12 July 2026 and Amazon Marketing Strategy: How They Built a Brand That Wins.
