It’s 2035. The highly anticipated opera, “A Dolphin Talks to a Boy” is set for its global premiere at Raleigh’s Meymandi Concert Hall. Its author is artificial intelligence visionary Igor Jablokov, making his debut as a librettist. Technology developed by Jablokov’s Pryon AI business composed the music. The show’s star, dolphin Alexa Siri, will perform in a
tank onstage.
Fantasy? Science fiction? For now, yes. But it will become a reality, says Jablokov, whose pedigree includes helping create AI voices for the very real Siri and Alexa software products relied on daily by millions of people.
“I remember as a boy in Greece seeing a dolphin that had been struck and injured by a propeller,” says Jablokov. “I asked, ‘Why can’t I talk to you?’”
Bottlenose dolphin Flipper had already starred in a TV sitcom from 1964 to 1967, before Jablokov was born in 1975. The fictional Flipper used chirps and squeaks to interact with humans. Jablokov’s dolphin encounter began a life-long ambition to explore science
and technology.
“My parents wanted me to be a part of the computer generation, so we moved to the United States when I was 4,” he recalls. “I didn’t even speak English.” His role models were his late grandparents in Philadelphia, who were very involved in his upbringing.
Given a Radio Shack computer complete with colored lights, Jablokov became enthralled and started “hunting and pecking away.”
His desire for a tech career was ignited, leading him to Penn State University, where he earned a bachelor’s degree in computer engineering while at times juggling three jobs to pay his tuition, room and board.
In 1996, Jablokov joined IBM in Charlotte. His projects included the company’s Watson supercomputer project, as well as voice technology used in an early PlayStation videogame system. He also earned an MBA at UNC Charlotte in 2000.
He spent just short of a decade at Big Blue before joining with his brother, Victor, to launch Yap!, where their digital-based interactive voice system became the basis of Amazon’s Alexa and Apple’s Siri. The former name is the same as Jablokov’s older sister, which he calls “just a coincidence.”
Yap! raised $6.5 million in 2008, backed by Sunbridge Partners and Harbert Investment Partners. Amazon also invested in Yap!, then acquired the business outright in a September 2011 transaction that drew little media attention. Victor, who has an MBA from Wake Forest University, is now CEO of Boston-based Blue Century, an augmented intelligence company.
Jablokov says he was set free by the sale and wasn’t interested in sticking with the Seattle-based e-commerce giant. Its interest in voice recognition technology surprised him. When considering potential acquirers, “Amazon didn’t even register” compared with IBM or Google, he says. But a “most lucrative” deal was inked, he adds.
NEW ADVENTURES
Post-sale, Jablokov served on some advisory boards at Penn State and UNC Charlotte, mentored several entrepreneurs and received an Eisenhower fellowship in 2013 and a Truman fellowship in national security a year later.
Mostly, he spent time studying artificial intelligence, including its expanded uses and possible safeguards to avoid potential harmful impacts. One morning upon waking from a dream, he immediately wrote down his recollections, which became the basis for his new company.
In 2017, Jablokov launched Pryon with a mission for what he calls “augmented AI.” Financial details remain private, but the business is attracting lots of interest in the business world’s hottest sector. Fewer than 15% of software startups achieve $10 million in annual revenue within a decade, but Pryon is ahead of that target.
“This company is evolving into something much more consequential,” he says. “[It] is going to be transformational. … We’re on a long-term journey here.”
Fewer than 10 of Pryon’s 100 or so employees work in Raleigh, with most of the rest based in the northeast U.S. While Jablokov is founder and chairman, he credits CEO Chris Mahl with helping turn the company from being “research and product driven to client-driven” since joining in 2021. After working for Oracle, Salesforce and other big companies, Mahl has led a series of smaller tech companies to successful exits over the past decade.
The chief technology officer is David Nahamoo, a 36-year IBM scientist who joined Pryon
in 2018. He understood the promise of cloud computing years ahead of the pack,
Jablokov says.
After initial fundraising rounds of $4.5 million and $20 million in its initial years, Pryon secured $100 million in September 2023. It expects to add more capital over the next year.
Several leading tech industry investors own stakes, including the U.S. Innovative Technology Fund, Breyer Capital and AOL co-founder Steve Case’s Rise of the Rest Seed funds. Early investors also included the Triangle Tweener Fund, Triangle Angel Partners and Carolina Angel Network.
Vice President J.D. Vance was working at investor Peter Thiel’s Mithril Capital in 2019 when the Austin, Texas-based firm invested in Pryon in 2019. “J.D. was a huge help. He led our Series A round and eventually connected us to others,” Jablokov says.
Vance served on Pryon’s board for two years before his U.S. Senate election in 2022. Jablokov says the two remain in contact.
“We’re definitely building for growth. We’re focusing on governance and how to construct things [whether it goes public or sells to a larger company.] We’re definitely hiring people with experience at public companies as well.”
Among Jablokov’s many admirers is veteran Triangle tech executive Scot Wingo, who leads the Triangle Tweener Fund. “What’s most impressive about Igor is his ability not to just see where the puck is going or see around corners, but how he can see five years into the future,” he says. When Pryon started, “it was in no way clear [AI] would be the huge success it is now.”
Pryon’s work was deemed among the “Top 10 Emerging Technologies” for defense and national security by consulting giant Booz Allen in a 2024 report.
“Igor is wired differently from most people. He’s an innovative thinker that isn’t afraid to speak his mind,” says Robbie Allen, a veteran Triangle tech entrepreneur involved with AI for years. “He’s been ahead of the curve on practical applications of AI as much as anyone I can think of.”
Jablokov wants North Carolina to become an early adopter of “transforming information into the intelligence age.” He is part of Gov. Josh Stein’s science advisory team, led
by Teena Piccione, a former Google executive who is the state’s secretary of
information technology.
STORYTELLER
No one can guarantee success, of course. The AI industry was rocked in late January after reports on the surprising popularity of China’s DeepSeek AI service. The news sent shares of AI leader Nvidia tumbling nearly 20% in a day. (It has since rebounded.)
Internet pioneer Marc Andreessen’s comment that DeepSeek was the “Sputnik moment for AI” didn’t impress Jablokov. “That’s absolutely overstated. Marc likes to post tweets. I prefer everybody study [issues] quietly, figure out the risks and opportunities.”
Jablokov likens Pryon’s strategy to that of Cisco and Apple. Cisco shifted from hardware to software, while Apple didn’t wait for a new Samsung device, instead inventing the iPhone.
“We’re going to have to be invested in small and large, not just at the cloud level,” he says. “We’ve always been striving to balance accuracy, security and speed since inception. We have
to be flexible. We have to operate at the nation-state level and also operate on an edge device that could literally be as small as your thumb.”
Pryon has developed what Jablokov calls a “first-of-its-kind knowledge operating system” that incorporates high-performance computing, natural language processing, speech recognition and other features that are mostly unintelligible to those without strong
tech skills.
Which brings us back to that mythical opera.
It would be a formidable task to incorporate dolphin sounds into what Jablokov calls a “knowledge fabric” of data, video, biologics and more. The result would be a Star Trek-like universal translator, making it theoretically possible to communicate with one’s dolphin
or dog.
“You may think that we will talk with aliens before we ever talk to animals, but that’s not true,” Jablokov says with conviction. “I believe this will happen within a decade.”
As for opera, the innovator compares it to his own life’s work. “AI is like an opera. It’s multidimensional. It takes us places where humans and STEM [science, technology, engineering, math] collide.
“The reason why AI is thrilling is that it is such a multidisciplinary field where everything matters, from computer and science to art.” It’s no wonder, he says, why the field is attracting trillions of dollars of investment “and some of the best talent on the planet Earth.”
Jablokov’s ability to pitch is one of his strengths, Wingo says.
“Igor’s a master storyteller. That enables him to weave a story that includes an incredible problem he’s going to solve and build a company around it,” he says. “You get J.D. Vance and top investors. Igor is also great at recruiting top talent using the same skill set.”
LIGHTS. CAMERA. ACTION.
Lyrics, dance, choreography, symphony, stages and sets, lights — the stage is ready. Jablokov is in a tuxedo and Alexa Siri has a microphone strapped to her chest. Together they introduce a new entertainment experience.
“I can finally talk with you,” Jablokov says. “I’ve been trying to talk to you my entire life!”
“You should have been swimming,” Alexi Siri responds.
“How can I help you?”
“I need stitches and an aspirin,” Alexa Siri says. “With a shrimp chaser.”
The bravos follow.
A LOOK UNDER PRYON’S HOOD
Igor Jablokov takes seriously the threat AI poses to the United States and the security of clients, and he’s skeptical of the Big Tech giants’ commitment to privacy. Protecting data is among Pryon’s top priorities, along with speed and energy efficiency. AI is a huge consumer of power, as noted by industry leader OpenAI’s plans to invest at least $100 billion in data centers to support AI applications.
Jablokov’s goal is for clients to use their own intellectual property and operations data to create custom AI models that improve productivity by streamlining processes and suggesting new products. Pryon is providing a suite of “knowledge management” software tools that serve a broad array of clients and run on devices ranging from large computers to mobile phones and other smart devices.
“We’ve been very careful in how we have built our businesses,” Jablokov says, emphasizing protection of intellectual property that is a contrast with the AI leaders. “[China’s]DeepSeek essentially pulled the same stunt as OpenAI, to suck in the bone marrow of creatives.”
But protecting creators’ rights is fundamental, he says. Why should somebody be able to suck [creative work] to make an AI model for some economic gain only for some?”
For competitive reasons, Pryon’s website doesn’t disclose all of the company’s innovations, Jablokov says. But the company has developed what it calls “ingestion and retrieval engines,” which various businesses and government agencies are using for accurate, verifiable information.
Examples of these engines are Generative Pre-trained Transformers (GPT), as seen in OpenAI’s ChatGPT to create human-like text, and Retrieval Augment Generation (RAG),
a process of optimizing a user’s question provided to an AI system. For example, if two people are trying to communicate but speak different languages, they use an interpreter (the “RAG”) to help translate for better understanding.
Retrieval Augmented Generation involves optimizing the output of “large language models,” an AI system that produces human-like text by referencing massive amounts
of data.
“The idea for Retrieval Augmented Generation was introduced in 2020, and I distinctly remember Igor talking about the basic concept when he was first talking about using natural language married with company databases to be able to ask an AI: ‘What are my sales on product X for the month’ type questions,” says Robbie Allen, who leads Durham-based Bionic Health, an AI startup. “The rest of the world is just catching up with this.”
Pryon and others are “only scratching the surface in terms of incorporating it into models,” he adds.
No matter the source or type of data, from handwritten notes to old photos, spreadsheets to video, Pryon’s technology produces intelligence in what Jablokov calls a “knowledge cloud.” Essential to success is producing accurate information. For example, a touted medical cure based on incorrect data, or an out-of-place decimal point, could lead to disastrous outcomes.
“Unquestionably, veracity matters,” says Tom Snyder, the executive director of RIoT, a Raleigh-based nonprofit that studies Internet of Things devices and applications. “AI trained on misinformation will have suboptimal results. AI trained on disinformation is even more problematic, as that is where bad actors are deliberately putting bias into AI, usually for malicious reasons like changing elections or harming healthcare.”
Pryon wants customers to compile information through its own enterprise data, rather than relying on potentially inaccurate sources.
“The ability as a company to marry the natural language power of [large language models] and a database of the company’s internal information with the assurance that the information won’t be trained on is going to unlock an amazing amount of productivity,” says Scot Wingo, who has led ChannelAdvisor, Spiffy and other Triangle tech firms. “We’re still in the early innings and Pryon is playing a big role in this space.” ■
