The pitch is almost too neat. Take a man convicted of securities fraud for running a boiler room, give him a voice AI company, and let him teach machines to close. That’s Tone, the new startup from Jordan Belfort, the real Wolf of Wall Street, and investor Shervin Pishevar. It’s easy to roll your eyes. It’s also easy to see why someone will write a check.
According to TechCrunch, Tone builds voice agents trained to use persuasive tonality. Belfort says there are about 15 tonalities that matter, and the model has been taught them one at a time, then taught to link them. Examples include turning declarative statements into questions to build engagement, and whispering to imply scarcity of information. The training data comes from his 30 years of sales coaching material.
The product is already in use. Belfort says customers include law firms, health insurers, polling operations, and auto businesses. In one demo scenario, the agent runs an entire health insurance sale: it asks qualifying questions about budget and ages, picks a plan, explains benefits, lowers the payment by raising the deductible, and swaps plans if the customer balks. Latency is around 600 milliseconds. To hit that speed, Tone uses its own small, light models rather than frontier models from OpenAI or Google, which Belfort admits costs some intelligence.
That admission is the most credible thing in the interview. He says a top human producer still beats Tone call for call, but that it should beat a mediocre closer, and he expects the gap to close within a year. Revenue is roughly $700,000 to $1 million, and the company has raised about $10 million from friends, family, and seed investors.
The new move is a self-service platform for authors and creators who want to turn books into expressive AI narration. Users can pick a Tone voice or, once voice cloning rolls out, use their own. This is a sensible side door. It puts the product in front of more users and likely feeds the training pipeline, which matters for a company whose stated edge is data quality.
Now the problems. The market for AI voice is packed. ElevenLabs owns mindshare in synthetic narration and cloning, while Bland, Retell, and Vapi sell conversational phone agents to the same sales and support teams Tone is courting. Tone’s differentiator is tonality tuned for closing, and that is a claim nobody outside the company can verify. Neither founder is an engineer, and TechCrunch notes it couldn’t independently confirm the technology works as described.
Then there’s trust. Belfort pleaded guilty in 1999 to securities fraud and money laundering, served 22 months, and was ordered to pay $110.4 million in restitution. Victims reportedly received far less than they were owed. Pishevar has faced misconduct allegations that he denies. Asked why a bank or hospital should trust them, Belfort said he made mistakes long ago and that some companies will pass while others will see the value.
Here’s why it matters beyond the spectacle. Persuasion-tuned voice AI in regulated sales, like insurance, raises real questions about disclosure, consent, and manipulation. An agent trained to whisper for scarcity while selling health plans will draw regulator attention sooner or later. Belfort says the model avoids hallucinating and lying, but that’s a promise, not an audit.
My read: the business is plausible, the technology is unproven, and the brand is both the best marketing asset and the biggest liability. Enterprise buyers in finance and healthcare will care about the second part. Smaller sales teams drowning in leads probably won’t. Watch the close-rate data, not the name.



