Jordan Belfort Launches AI Voice Startup Tone, Aiming to Sell Like the ‘Wolf of Wall Street’

6–10 minutes

From Wall Street Scandal to AI Entrepreneur

Jordan Belfort, the self‑proclaimed “Wolf of Wall Street,” spent 22 months in prison after pleading guilty in 1999 to securities fraud and money laundering tied to Stratton Oakmont. A court ordered him to pay $110.4 million in restitution to 1,513 victims, a sum that ultimately fell short of what was owed. In late April of this year Belfort completed his restitution obligations, yet he continues to draw attention for his relentless reinvention.

Recent years have seen Belfort pivot to sales consulting and host cryptocurrency workshops in his Miami Beach home, where admission once cost one bitcoin or its $40,000 equivalent at the time. Now he is betting that his expertise in persuasion can be encoded into artificial intelligence, launching a voice‑based platform called Tone.

The launch echoes an earlier episode in tech investing: years ago, as doubts grew over WeWork’s sky‑high valuation, I interviewed co‑founder Adam Neumann. Despite SoftBank’s lofty numbers, investors were eager to back him. The same willingness to suspend skepticism appears to be fueling interest in Belfort’s new venture, even as critics note his co‑founder Shervin Pishevar’s own controversial past.

Tone: A Voice Platform Built on Persuasion Science

Tone is designed to embed the tonality that Belfort has refined over three decades of teaching sales and influence. The platform’s “secret sauce” is a set of roughly 15 distinct tonal patterns—such as phrasing declarative statements as questions to build engagement, or whispering to imply information scarcity—that Belfort says drive higher close rates. He calls this approach “goal‑oriented communication,” where the way a message is delivered matters as much as its content.

The technology began as a personal project: Belfort trained a model on his own voice and 30 years of high‑quality audio, achieving a synthetic voice that matched his tonal signature. He then layered additional tonalities and taught the system how to transition between them. The co‑founder, longtime investor Shervin Pishevar, joined after Belfort realized existing AI voice tools lacked the nuanced delivery he needed. Together they built a proprietary system, with VP of engineering Ben Sussman overseeing the technical rebuild from Belfort’s prototypes.

Beyond its core sales application, Tone is rolling out a self‑service platform for authors and content creators who want to turn books into expressive AI‑generated narration. Users can select from a library of AI voices or, once the voice‑cloning feature is widely available, create a clone of their own voice—provided they speak into the system and sign a consent form.

How Tone Works and Who Uses It

Tone is already deployed by law firms, health insurance providers, polling organizations, and automotive companies. The platform can shepherd a prospect through an entire sales cycle: it asks qualifying questions about coverage needs, budget, and demographics; selects a suitable plan; explains features and benefits; and can even suggest adjustments to lower premiums or switch plans if the customer objects. The system operates with a latency of roughly 600 milliseconds, allowing near‑real‑time interaction.

When a conversation deviates from the script, Tone is engineered to avoid generic filler such as “That’s a great question.” Instead, it addresses the unexpected query, maintains rapport, and smoothly returns to the intended flow without hallucinating facts. The model has been trained to recognize the boundaries of a “happy path” while respecting the buyer’s input.

Revenue, Funding, and Market Position

Belfort reports that Tone has generated approximately $1 million in revenue, about $700,000 of which came from client contracts. The company has raised $10 million in seed and friends‑and‑family funding. While the market is crowded with voice AI players such as ElevenLabs and Cartesia, Tone’s differentiator is its patented tonality‑intelligence layer, which decides when to apply a specific vocal style. Belfort argues that sounding human will soon be a baseline feature for all competitors, but the strategic timing of tonal shifts is what gives Tone an edge.

The regulatory landscape adds another layer of complexity. The FCC ruled in 2024 that AI‑generated voices fall under the Telephone Consumer Protection Act’s robocall rules. Belfort insists Tone only contacts opt‑in leads or existing customers, avoiding cold calls that could trigger TCPA lawsuits. Polling is the sole exception, where an exemption permits calls to private households without prior consent.

Voice cloning is handled with strict consent mechanisms. To create a clone, a user must speak live into the system, confirming identity and agreeing to the terms. Uploading pre‑recorded audio is discouraged; if attempted, the platform requires a real‑time voiceprint match. These safeguards aim to prevent impersonation and ensure that cloned voices are used only with the original speaker’s permission.

The Road Ahead

Belfort says he plans to raise a “significant amount” of additional capital to fund compute reserves and further product development. While he is open to a substantial acquisition offer, he currently intends to keep Tone independent. The company’s pricing model blends per‑minute charges for streaming, monthly subscriptions for the content platform, and success‑based fees for select clients.

This summer Paramount+ released a documentary about Belfort, but he has not watched it. He avoids reading commentary about himself, preferring to follow his “internal compass.” Whether investors will follow his lead into the AI space remains to be seen, but his confidence in the technology—and his willingness to pitch it directly—suggests the conversation is just beginning.

Interview with Jordan Belfort

What have you built? I created a voice AI that applies the same tonality principles I’ve taught for decades. It’s not just about replicating a voice; it’s about delivering the right vocal cues at the right moments to influence a buyer.

Is this effectively trained on you? We started by training on my own voice and 30 years of high‑quality audio. The model captured my tonal signature, then we added a second tonality and taught it to link them. Data quality is everything—any corruption can undo progress.

How did you and Shervin Pishevar come together? We’ve been friends for years. After hearing a disappointing AI sales demo, I realized we needed a voice that could truly teach selling. Necessity drove us to build our own solution.

Who are your customers, and how do they use it? Law firms, health insurers, pollsters, and auto dealers use Tone to run full sales cycles. It asks questions, proposes plans, adjusts pricing, and can switch options—all within about 600 ms.

How does the AI handle a sales call that goes off script? It never defaults to generic phrases. The model addresses unexpected questions, keeps rapport, and returns to the intended flow without hallucinating, ensuring the conversation stays honest and on track.

Have you generated any revenue from Tone yet? Have you raised funding for it? Revenue is roughly $1 million, with about $700,000 from clients. We’ve raised $10 million from friends, family, and seed investors.

You have a fraud conviction. Pishevar has faced misconduct allegations. Why should a bank or a hospital trust this product? I acknowledge my past mistakes, but for the past two decades I’ve taught sales and persuasion worldwide. Some companies may opt out, but many see the value in what we’ve built.

What close rate does Tone have versus a typical sales force? It’s not yet as strong as a top producer, but it outperforms mediocre salespeople. We expect improvement within a year as the model matures.

Where is it used most effectively? Low‑dollar sales opportunities? The platform shines where lead volume exceeds human capacity, across time zones or languages. Great human salespeople will always be needed; Tone is a force multiplier, not a replacement.

You’re competing with ElevenLabs and Cartesia. Tone is your edge? Our edge is the patented intelligence that decides when to use a specific tonality. Sounding human will become standard; the timing of vocal cues is what sets us apart.

The FCC ruled in 2024 that AI voices fall under robocall rules. How does outbound sales calling work? We only call opt‑in leads or existing customers. Cold calls are off‑limits to avoid TCPA lawsuits; polling is the sole exemption.

How do you verify consent and prevent impersonation in voice cloning? Cloning requires live speech and a signed consent. If someone uploads audio, we’d need a real‑time voiceprint match. We didn’t launch cloning yet, but the safeguards are in place.

Who created the technology? You’re not an engineer, and Pishevar isn’t, either. I saw a name on the patent, Jordan Anthony Metzner. Ben Sussman is our VP of engineering and leads the technical work. I prototype ideas, and my engineers rebuild them. The patent does list Jordan Anthony Metzner, but he is not the engineer behind the product.

Do you charge per minute, or based on the outcomes customers see? It’s a mix: per‑minute for streaming, monthly subscriptions for the content platform, and success‑based fees for some clients.

Is this an independent company or something that belongs inside a larger AI company? For now Tone is independent. If a substantial offer came that made sense, we’d consider it, but my plan is to raise more capital and keep building.

Paramount+ released a documentary about you this summer. Did you watch it? I didn’t watch it. I avoid reading or watching anything about myself, good or bad. I follow my own internal compass.