Belfort Acknowledges the Limits of Its Commercial AI

Jordan Belfort develops Tone, a voice platform that he claims can drive sales through intonations. The company is also launching a narration service for authors and creators. Its commercial performance remains based on assertions from its co-founder, who acknowledges that the best human salespeople still outperform it.
Belfort presents still uncertain revenues
About one million dollars, or perhaps 700,000 dollars: Jordan Belfort gives a hesitant estimate of Tone's revenues. However, he announces that he has raised around 10 million dollars from close contacts and during a seed funding round. He now plans to seek a significant amount, without specifying the figure.
Billing depends on usage, he explains: streaming is charged by the minute, while the content offering is to be provided in the form of monthly subscriptions. For some clients, he believes that most of the revenue relies on success-based compensation. Belfort does not rule out discussions about a buyout if a sufficiently high offer seems relevant; he mainly states that he needs available computing power, without deeming it necessary to own data centers.
Regarding commercial effectiveness, he admits that an excellent salesperson currently surpasses Tone on a call scale, while believing that this ratio could change in six months to a year. He also estimates that his tool will significantly outperform a mediocre salesperson tasked with closing deals. Speed, according to his explanation, requires small, lightweight proprietary models rather than cutting-edge models: he acknowledges that they have lesser intellectual capabilities.
Tone targets prospects that teams cannot handle
Belfort cites law firms and health insurers among its users, as well as activities in polling and the automotive sector. He presents their demand as a lack of personnel in the face of the volume of prospects, but also as a need for availability across multiple time slots or in several languages. He claims not to seek to replace humans and believes that the best salespeople will retain their usefulness.
In health insurance, he describes an agent capable of asking for the desired coverage, budget, and age, then selecting a plan and detailing its benefits. If the payment seems too high, the system could explain how a higher deductible would reduce it; if the plan is unsatisfactory, it would present another option. Belfort thus claims to handle the entire sale, with a response time of about 600 milliseconds.
He also states that he has trained the model to address the prospect's concerns before resuming the planned sales journey, without degrading the relationship or producing fabricated information. Systematic compliments on the questions asked are, he says, discarded: he finds them too flattering. These descriptions do not constitute results from independent tests.
Belfort bets on the choice of intonations
Against ElevenLabs or Cartesia, Belfort claims a difference that lies less in the realism of the voice than in the selection of the appropriate intonation. He believes that competing voices will all sound human in six months to a year and asserts that he holds a patent on the choice of tonalities. He claims that those from ElevenLabs and Cartesia did not meet his needs.
Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisissez votre rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
His method stems from a teaching of persuasion that began in the late 1980s. Belfort identifies about 15 particularly effective intonations and describes communication guided by the desired outcome, for example, appearing as an expert. In his view, a statement pronounced as a question elicits engagement, while a whisper suggests information that is not easily accessible.
He recounts that he initially wanted to teach sales to models after hearing a commercial AI demonstration that he found very poor. Lacking a satisfactory voice, he says he set out to build one, starting with his own based on 30 years of data that he qualifies as very good. The training would have focused on each tone separately, then on their sequences.
Belfort considers the reproduction of his voice and intonations to be accurate; he also claims that a small volume of corrupted data can significantly compromise progress. These assessments remain his own. The operation of the technology as described by the founders has not been independently verified.
Tone prepares cloning for its narration offering
The opening to content targets authors wishing to have their books narrated and creators seeking expressively generated artificial narration. The self-service platform provides access to the voices offered by Tone. This activity could also further feed its training.
User voice cloning is in preparation, but Belfort assures that it has not been launched, although he believes it is ready to be. He describes a process that requires speaking in front of the computer and signing a consent form. In the case of an imported audio file, a practice he advises against, he plans to verify the match with a voiceprint captured live.
Belfort claims to prohibit cold calls
In 2024, the FCC subjected AI-generated voices to the automated call regime outlined in the U.S. Telephone Consumer Protection Act. Belfort claims that Tone restricts its commercial calls to prospects who have consented to be contacted or to existing customers. He states that he prohibits the use of the tool for cold prospecting, citing potential lawsuits under the TCPA.
However, he presents polling as a distinct case. According to him, an exemption allows contacting private households without obtaining prior consent.
Founders must account for their backgrounds
The platform brings together Belfort and investor Shervin Pishevar, whom he describes as a longtime friend. Shervin Pishevar has faced allegations of inappropriate behavior, which he contests. Jordan Belfort, who has dubbed himself the Wolf of Wall Street, pleaded guilty in 1999 to securities fraud and money laundering in connection with the Stratton Oakmont case, a Long Island brokerage involved in low-value stock manipulations.
He served 22 months in prison and was ordered to pay $110.4 million to 1,513 victims. In July, TMZ reported that they received significantly less than the amounts owed, while also noting the completion, at the end of April, of Belfort's court-ordered restitution obligations. His autobiography was adapted into a film by Martin Scorsese titled The Wolf of Wall Street.
Now a business consultant in Miami, Belfort has also hosted workshops at his home in Miami Beach focused on cryptocurrencies. A New York Times journalist attended a weekend workshop priced at one bitcoin, which was $40,000 at the time. When asked about the trustworthiness of Tone, Belfort acknowledges past mistakes and cites the last 20 years spent teaching sales and persuasion to companies worldwide; he admits that some may refuse to work with him.
None of the co-founders are engineers. When questioned about Jordan Anthony Metzner, whose name appears on the mentioned patent, Belfort identifies Ben Sussman as the engineering lead. He describes his own contribution as evaluating the sound output and creating prototypes that engineers then completely rebuild.
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.