Should You Build an AI Clone of Yourself?
Somebody has probably already pitched you on this. If not, they will.
There’s now a full industry selling coaches, consultants, authors and course creators an AI version of themselves — trained on your books, your talks, your frameworks, your way of saying things. It answers your audience at three in the morning. Some people charge a thousand dollars a year for access to theirs.
I’ve been sitting with this for a few weeks. Three books, a decade of frameworks, a lot of recorded thinking — on paper I’m exactly who these platforms are built for. So this is partly me working out my own answer in public. Here’s where I’ve landed.
Clone the answers. Never clone the judgment.
Why It Matters
The pitch is genuinely appealing, and I want to give it a fair hearing before the cautions.
Most people selling expertise answer the same eight questions forever. A clone handles that endlessly, in any time zone, without you. It gives people who’ll never afford your rate a real taste of how you think, and it supports course students at the moment they get stuck — which is the moment most of them quit. For the repetitive half of an expertise business, it mostly delivers.
The trouble starts when the clone crosses from answering into advising.
Tomorrow’s forecast, in your inbox.
One email. Five minutes. Written for owners, not engineers.
What to Clone and What to Keep
This is the part I’d write down before looking at a single platform.
| Safe to clone | Keep for yourself |
|---|---|
| Frequently asked questions | Strategy for a specific business |
| Your frameworks and how they work | Negotiation |
| Onboarding and getting started | High-stakes diagnosis |
| Course and program support | Final recommendations |
| Knowledge retrieval from your material | Original thinking on a new problem |
| Qualifying whether you’re a fit | Anything with real money or real risk attached |
The left column is retrieval — someone needs a thing you already know and have already said. A machine does that well, at four in the morning.
The right column is judgment applied to one person’s situation. That’s what people pay premium prices for, and it’s what a clone does confidently and badly, because it will answer fluently whether or not it should.
What It Actually Costs
As of July: Personify has a free tier with paid plans from $39 a month, Coachvox is $99 flat, Delphi runs $99 to $349, BuddyPro is $197. Custom agency builds start around $10,000 and run well past $100,000. But the number that matters isn’t on any pricing page.
Check the revenue share before you check the monthly fee. If you charge people for access to your clone, some platforms take a cut. Coachvox takes 10% per transaction plus card fees, by its own FAQ. Delphi’s last publicly documented rate was 20% of subscription revenue, and its current terms are negotiated privately. Personify and BuddyPro take nothing.
Two platforms that both look like a hundred dollars a month are not the same deal. On a clone earning $100,000 a year, a 10% share is $10,000 annually — several times the platform fee, invisible at signup.
A Note on Who’s Telling You This
I went looking for an independent comparison of these platforms and couldn’t find one. Every substantial guide in this category is published by a company selling a competing product, each ranks itself first, and almost none say so. The most neutral-looking source I found is a consultant who builds AI twins for a living.
So, before my recommendation rather than after it: we earn a commission from one of the platforms named here. It didn’t soften the revenue-share warning I just gave you about that same company. Interest first, recommendation second — that order is the whole difference between this page and the ones ranking above it.
Read their numbers carefully too. One platform advertises that coaches reach six-figure recurring revenue within thirty days, and elsewhere reports 150 experts producing $5 million in total. Divide it out and the average is about $33,000 per expert, lifetime. Both figures are theirs.
Doing It Without Losing the Thing You’re Selling
Three rules I’d hold myself to, and they’re the reason I haven’t built one yet.
Say what it is. Nobody should wonder whether they’re talking to you. Half of consumers already prefer businesses that keep AI out of what they see, and getting caught pretending is far more expensive than disclosing. Name it plainly on the page.
Don’t train it on things that aren’t yours. Your books, talks and published material are fair game. Client session transcripts belong to your clients, whatever your terms of service technically permit.
Watch what it does to you. This is the one I keep circling. A clone is trained on who you were when you fed it. If it takes over the repetitive work, you get time back for harder problems, and your thinking keeps developing. If it takes over the client contact, you stop encountering the situations that made you worth cloning in the first place. Your expertise stops updating, and the clone slowly becomes a recording of someone who no longer exists.
The difference between those two outcomes has nothing to do with the software. It’s about what you do with the hours it gives back.
Who Should Pay Attention
This makes sense if you have real depth of published material, an audience larger than your calendar, and questions you answer over and over. It doesn’t if your material is thin — a clone trained on four blog posts sounds like a chatbot wearing your name, which is worse than nothing — or if your value is bespoke judgment for a handful of high-paying clients. You’d be automating the wrong half.
The Owner’s Forecast
Within two years, an AI layer on top of expertise becomes ordinary — the way having a course became ordinary. Most will be free, sitting on the website answering questions, and the ones charging real subscription money will be the exception rather than the model.
The split I’d expect: experts who used it to remove repetition will be visibly sharper, having spent the reclaimed hours on harder problems. Experts who used it to remove themselves will be running a business built on a snapshot of their thinking from two years earlier, and their audience will feel the staleness before they can name it.
Moderate rather than high, because the platforms are consolidating fast and today’s pricing structures are unlikely to survive intact. The capability isn’t going anywhere.
Your Next Move
Good — 20 minutes. Write down the questions you answered more than three times last month. That list is your honest case for a clone, and for some people it’s short enough to end the conversation there.
Better — an afternoon. Before paying anyone, build a rough version inside whichever AI tool you already use. Load your material, ask your ten most common questions, and read the answers as a prospect would. You’ll learn more in an hour than any comparison page can tell you — including whether you have enough material to be worth cloning.
Best — ongoing. If you go ahead, check the revenue share before the monthly price, disclose it plainly, and decide in advance which questions it hands off to you. Then protect the reclaimed hours for the judgment work. That last part is the whole return on this, and it’s the part nobody enforces but you.
Today’s Instrument
Coachvox
The starting point if what you want is the left column
If the case is there, Coachvox is where I’d start. Built for coaches rather than public personalities, one plan at $99 a month with unlimited conversations, and it hands you transcripts of what people actually asked. That last part is worth more than it sounds — a running record of what your audience is confused about.
- Best for
- Answering the same questions endlessly, qualifying whether someone’s a fit, and supporting course students at the moment they get stuck. The retrieval work in the left column above.
- Skip if
- You intend to sell subscription access to your clone. Coachvox takes 10% per transaction plus card fees, which is the cost I flagged earlier and I’m not going to pretend otherwise now that I’m recommending them. At real volume a flat-license platform costs less. Also skip it if your published material is thin — no platform fixes that.
Almanac note: this link may earn us a commission at no additional cost to you. It didn’t soften the revenue-share warning above, and it never changes the verdict.
Owners Are Asking
Do I need a separate platform, or can I use ChatGPT or Claude?
For testing, use what you have. The dedicated platforms add what makes it a product rather than an experiment: a public interface, memory across conversations, lead capture, and billing.
Sources: Platform pricing verified against live pricing pages, July 2026 · Coachvox published FAQ on transaction fees · Gartner Marketing Survey, March 2026