In the last twelve months, the technical possibility of building a digital clone of a person, voice and face, starting from a few minutes of reference video, has become accessible to anyone. There are public services that do it for a few euros a month. There are TV presenters who have their own avatar, there are CEOs of multinationals who send Christmas greetings to their employees in fifty different languages using a clone, there are marketing gurus who generate their content for Instagram from a script and never film themselves again.
For about a year, even much smaller companies, particularly in distribution, real estate, retail, have been asking us if it makes sense to do it for them. The answer we give is not yes or no. It is: it depends on three things, and the three things must be kept separate. Mixing them is how you end up using a powerful tool in the wrong places, obtaining the opposite of the intended result.
In this article we try to separate the three things.
Three rules, in order of importance
First rule: the clone is built with the explicit, written consent of the person being cloned, on content that remains under the control of the company or the person themselves. This is the rule with no exceptions. You do not clone a person without them knowing and having signed. You do not clone a person to produce content the person would not have approved.
It seems an obvious rule, and it is not. The temptation to "clone the founder quickly because he is always away" without having him sign anything, or "to make him say something he would have refused to say in person", is real and quite widespread. When it happens, the reputational damage is enormous and immediate, and even legally the position becomes fragile. The rule of explicit consent is therefore the first one for ethical reasons, and the first one also for pure operational defence.
Second rule: the clone is used for brand-owned content, that is content coming from the company or the person themselves. Editorial videos, posts on company social media, internal communications, commercial materials. Never for testimonials of clients, employees, patients, third parties who have not been cloned. A digital clone that looks like your CEO is fine; a digital clone that looks like a customer telling their experience, no. Never. The difference between the two things, seen from the outside, is the boundary between a legitimate tool and manipulation.
Third rule: the clone is not used in sectors where professional credibility is capital. Medicine, trustee work, legal, financial advice, healthcare in general. In these sectors, the professional who speaks is the professional. Their physical presence in the video is part of the service, not a costume. A doctor who produces educational content on a treatment, speaking through their clone, is creating a hybrid that, if discovered, destroys the trust the patient has placed in them. The risk is asymmetrical: the efficiency gain is marginal, the reputational loss is structural.
Where instead clones do change things
The three rules take most of the reckless uses off the table. What remains, however, is a space where digital clones solve real problems and produce measurable value.
The clearest case is that of executive figures in distribution, mobility, premium retail, real estate companies, where the public presence of the person counts a lot and that person's time is chronically saturated. The commercial director of an automotive dealer who would like to publish a video a week on social media, with their face, but manages to film one a month for lack of time. The founder of a chain of jewelleries who would like to produce content for LinkedIn, knows it would work, but under operational pressure gives up.
For this profile, the digital clone, built with the explicit consent of the person, on a script the person approves, allows producing coherent editorial series without saturating the person's time. It is not a substitute for real presence: it is an extension that allows maintaining a cadence one would never maintain alone.
The second useful case is the translation of video content. An executive figure films a video in Italian. The same video, dubbed by the clone in English, French, German with the same voice and the same timbre, reaches markets the original language would not have reached. For a company with cross-border clientele, this changes the economics of content production.
The third case is the generation of micro-content at volume. Replies to LinkedIn messages recorded as short personalised videos, each reply different, in two minutes without the person touching a camera. For those handling high volumes of interaction, it is a quality of contact that without a digital clone is simply unreachable.
What is needed to do it well
Making a digital clone work for an executive figure requires a technical chain and a process that cannot be improvised.
The technical chain includes the reference recording session (one to two hours of clean video, in multiple angles, with consistent diction), the voice and visual cloning model (some commercial platforms today produce results indistinguishable in internal corporate use), the generation pipeline (script, voiceover, video generation, quality control). Nothing insurmountable, but every step has pitfalls.
The process includes three oversight steps often skipped. The written and periodically renewed consent of the person, with precise definition of what the clone can and cannot do. Human revision of the script before generation, because a digital clone saying something incorrect is still something that goes out with the voice and face of the person. The conservation of reference material and the log of generations, in order to be able to reconstruct afterwards what was done by whom.
A technical note that affects the choice of supplier. Commercial cloning platforms leave a recognisable imprint, and this imprint is becoming, week after week, more detectable by moderation systems and attentive observers. A clone built in-house, on controlled infrastructure, almost always has a more discreet rendering. For internal and brand-owned use cases, the difference does not show. For cases that go out publicly at high volumes, it counts.
The digital clone for executive figures is one of the tools that today allows producing more content, more quickly, while maintaining a coherent voice. The difference between a use that works and a use that causes damage lies entirely in the three initial rules, applied without shortcuts. Without these three rules, the tool becomes what the sceptics suspect: another way of selling smoke at low cost.
If your company is evaluating a path of this kind, we talk about it in a forty-five-minute conversation. The diagnosis is free, and even before the clone we talk about the voice.