Natalie Ashes on why you clone your best people instead of replacing them
“The only scary thing about AI is someone that knows how to use it better than you.”
The episode
Natalie Ashes was a CMO when she built a clone of herself. She called it Marketing Nat and loaded it with every case study and lesson from her career. Within about a month she says it was better than her, smarter than her and more strategic. So she went and learned conversational AI with a whiteboard marker on her bathroom mirror, and now she runs WorkClone, an Australian company that clones a business's best people, from their voice to the way they work. Her team is about 10. She says the same build six years ago would have taken 100 people and two to three years.
James and Natalie get into what a clone is and how it differs from an off-the-shelf agent, why the real person has to stay, and what happens when a CTO decides to build voice AI in-house. She is direct about where it goes wrong. An agent that quotes couch cushions as a bedtime story. A model that gets the maths right 98% of the time in an industry where 98% does not cut it. They finish on the hires an AI company makes first, the go-to-market role growing out of the BDM, and how far behind Australia is.
What we get into
- Clones amplify a team. They do not replace it. The real person has to stay, because without them the clone is stuck at a point in time while the scripts, the business and the market keep moving.
- Natalie does not start with a six-week consulting engagement. An internal tool called the Clone Father reads the business, ranks the five or ten clones worth building by ROI and speed to market, and the first build is capped at one or two workflows.
- She rang her couch company for a quote and got an off-the-shelf voice agent. Asked for the quote as a bedtime story, it said no. Asked again, it started telling one. That is what an agent without proper guardrails does with your brand.
- 98% is not good enough. One WorkClone payment clone is really 44 sub-clones, with plain old code handling the dates and the maths because a model cannot be trusted with either.
- Who owns your clone when you leave? Natalie's view is that your voice and personality go with you, and the workflows and procedures stay with the business.
- WorkClone hired technical people first and still has no salespeople. Natalie is the sales team. The go-to-market role everyone is chasing is what the BDM has turned into: technical enough to scope the work and talk to the dev team, and able to hold a room.
- Natalie says her team surveyed 1,000 executives and around 88% admitted to feeding private company data into AI models. Flicking the switch that says do not train the model does not make that data secure.
Lines worth keeping
Tap to copy, then drop it into a post.
“Every business, most businesses, have a few unicorns. They're just people that are more commercially effective. And they're also really hard to replace. So we kind of hone in on those unicorns and we clone them.”
“98% of the time, it will get the calculations right. And 98% of the time, it will tell you that the call is being recorded. But that's not 100% of the time. When it comes to certain guardrails and certain industries that we play in, 98% is not good enough.”
“I feel that when someone leaves, their personality should go with them, but not the operational procedures underneath it.”
“What used to take four weeks to do an API integration now takes four hours, but we still need someone that's driving the ship.”
“Just because you can build doesn't mean you should. The level of detail that you're going in, especially if your customer is going to use this, is the reason that you shouldn't be building it yourself.”
Why it matters
Every board is asking how many people AI saves. Natalie runs an Australian AI company with about 10 people where she says it once would have taken 100, and her answer is still that no role is 100% cloneable.
Chapters
Tap any chapter to jump to that moment in the video.
- Nat Clone previews the episode
- What WorkClone does: cloning your unicorns
- Start small: the Clone Father and one or two workflows
- Build or buy? When the CTO says "we'll do it ourselves"
- The couch company and the bedtime story
- One clone, 44 sub-clones and why 98% isn't good enough
- Cloning real people: jobs, IP and personality
- From CMO to AI founder
- Careers, creatives and the mid-level squeeze
- First hires and the rise of go-to-market
- Is Australia behind on AI? The air conditioning test
- The team maths: 10 people vs 100
- Do clones say they're clones?
Headcount & Code
Get the read that goes with the show
One issue every week. The moves that matter in Australian tech hiring, for the leaders making the hire and the tech professionals making the move.
