Voice AI researcher and cross-disciplinary engineer on how latency, not linguistics, will define the interface revolution
According to a recent market analysis, the AI in the voice assistant market is projected to grow from $3.54 billion in 2024 to $4.66 billion in 2025, with 8.4 billion voice assistant devices expected to be in use worldwide by 2025. Yet, voice remains underused in enterprise environments and business automation.
What holds it back, and what’s about to change? AI Time Journal speaks with Vitaliy Danylov, co-founder of a U.S.-based voice AI startup focused on cross-border communication. Danylov holds two master’s degrees (from NYU and DNU), has authored a book and three peer-reviewed papers on voice AI, and previously developed enterprise-grade solutions for companies like Take-Two Interactive Software, Shiloh Industries, and Tower International. In 2025, he served as a judge for the 20th Annual Globee Awards for Technology, evaluating over 50 submissions in the areas of AI and cloud infrastructure.
“People tolerate robotic tone more than they tolerate five-second delay”
Vitaliy, most specialists come to voice AI from purely technical fields. You have a rare combination: financial analytics, political science, and now computer science. Does this understanding of business, human behavior, and technologies give you a special vision of why voice will become the dominant interface?
Yes, my background gives me a unique lens. Finance and business analytics taught me how businesses think, what technologies stick and what don’t. Political science and other social science classes I took provided me with insight into human behavior: what people adopt naturally and what feels forced, regardless of how well it’s marketed. And my tech experience lets me assess what’s implementable. That three-angle view helps me filter out hype. Voice is faste, at least 3 times faster than typing, and for the first time, speech recognition is accurate enough to handle real-world noise, accents, and latency. That tipping point just happened recently, and it’s why I believe voice will start replacing text in many human-machine interactions. As voice AI becomes fast and stable enough for production environments, it naturally merges with another trend: the rise of AI-powered digital workers. What used to be a chatbot becomes a full digital agent — capable of listening, reasoning, and respondin, in natural speech.
Drawing on your Master’s in Financial Management from New York University, how do you assess the financial rationale behind replacing office workers with voice-enabled digital employees?
White-collar roles often come with higher base salaries and bonuses. If you can automate those functions, the ROI is visible immediately. Investors and CFOs model this with a simple equation: Is the present value of the expected gain, that is, reduced expenses plus increased revenue, worth the predicted risk, which is the cost of failure multiplied by the likelihood of failure? When the answer is yes, automation proceeds. When it’s no, humans stay in the workflow loop. There’s also a risk exposure angle. When a digital employee makes a mistake in customer support, it can, in the worst case, mildly frustrate someone. However, if a digital employee discusses the legal case with the wrong client or authorizes a miscalculated vendor payment, the legal or financial exposure can be substantial. That changes the math. So, in practice, we’ll see digital employees enter office roles first, where the work is high-cost, low-variance, low-risk, and scalable. Everything else will lag, not because it can’t be automated, but because the numbers don’t justify it — yet.
“Voice creates 5x more input — and provides more environment context”
Based on your experience working with enterprise systems at companies like Take-Two Interactive Software, valued at $28 billion, and Shiloh Industries, where you implemented solutions for 25 global automotive plants, how do you see voice interfaces integrating into corporate environments?
In an enterprise, tech gets adopted when it either cuts costs or increases revenue. Voice does both. It can augment or replace human agents in high-cost regions, provide 24/7 support without wait times, and eliminate the need for call rerouting on holidays or weekends. On the revenue side, think about car dealerships — over half of inbound calls go unanswered. That’s lost sales. A voice agent handling those calls, even with a modest conversion rate, can make a difference. My experience with large-scale enterprise systems has shown me that when a technology becomes fast, cheap, and stable enough, it stops being futuristic and starts being deployed. Voice is right at that threshold. But to make voice-based digital employees viable at scale, cloud infrastructure has to catch up.
In your startup, you’re developing scalable cloud technologies to help cross-border businesses communicate more efficiently using AI voice systems. How does cloud computing architecture affect the speed of voice technology adoption?
Voice tech sits between text and video in terms of complexit, it’s lighter than video streaming, but much heavier than typing. Processing audio in real time requires serious cloud muscle, and latency adds up fast if services are scattered. The most effective systems put ASR, LLMs, and TTS in the same physical instance or data center. If you’re hopping between clouds, delays become visible. That’s why the best cloud providers — AWS, Azure, Google Cloud — aren’t just fast; they’re integrated. They offer things like sentiment analysis and translation under one roof. Voice tech adoption will scale fastest where the architecture minimizes friction for developers.
“The winning business models will mirror human employment.”
As a co-founder of a startup, you understand market dynamics from the inside. What business models will become dominant in the digital employee space? Subscriptions, licenses, or something fundamentally new?
I think the dominant models will be subscriptions and performance-based transactions, depending on the use case. The subscription model will be the default, especially for internal support roles — customer service, reporting, and task automation. You’ll pay a flat monthly fee, just like you pay a human salary. It’s easy to budget, easy to compare, and aligns well with existing workflows. If the digital employee replaces a $6,000/month office role, and the bot costs $600/month, that’s an easy sell. Transactional models will gain traction in performance-based functions, like sales bots. There, you might pay a percentage of revenue generated. It’s similar to how contingency-based lawyers work: they only get paid if they deliver. That model is risky for vendors, but incredibly appealing to buyers.
The winning model will be the one that mirrors human employment most closely. The subscription mirrors payroll, and the transaction model closely resembles work for commissions. That framing will help companies onboard digital employees without rewriting their entire mental model of work.
Your experience migrating financial systems for 25 global automotive plants showed how fast digital transformation can happen at scale. What lessons apply to deploying digital employees?
One of the biggest lessons I learned is that you can’t automate what isn’t documented.
Human workers can make educated guesses, adapt in real time, and connect the dots when something is missing. Digital employees can’t. If a workflow isn’t fully mapped out, with all its inputs, outputs, exceptions, and failure cases, you risk hallucinations and breakdowns that no one notices until it’s too late. If your instructions are unclear or your business logic is buried in years of hard-to-describe internal knowledge, you’re not ready for automation, no matter how powerful the underlying process automation model is.
Also, trust matters. Just like new human employees, digital ones have to earn their place. You don’t give them mission-critical tasks on day one. You start small, observe closely, and onlythen scale them across geographies or business units. That mindset, slow onboarding, fast scaling, is critical for digital transformation to work.
“Even among top AI startups, voice is still seen as niche.”
As a judge for the 20th Annual Globee Awards for Technology 2025, evaluating 50 submissions in AI and cloud categories, what trends in voice technologies do you observe among modern startups and corporations?
What stood out is how little attention voice tech is getting, even among cutting-edge startups. Out of 50 submissions I judged, maybe 2 or 3 were truly focused on voice. Most were centered on text and LLM-based workflows. That tells me voice is still considered niche, even though it offers massive gains in speed and usability. I think part of the hesitation is financial, venture capital tends to fund what’s trendy, and voice hasn’t hit that peak yet. However, I believe it’s exactly in these overlooked areas, such as voice and vision, that the next big leap will occur. Humans are wired for speech; adoption is just a matter of infrastructure catching up. The shift from text to voice isn’t just technical. It’s cultural, and generational. I see this firsthand mentoring NYU students.
“The next billion users won’t type — they’ll speak”
As a mentor in the NYU Alumni in Tech Club, what skills do you recommend young professionals develop to be ready for the era of voice technology dominance?
When NYU students ask me how to future-proof their careers, I tell them it depends on where they are. If you’re early in your career, stay curious and flexibl, learn broadly and explore fast. If you’re more experienced, specialize and go deep. As for voice tech, it’s not about learning “voice skills”, it’s about realizing voice is just another input. LLMs are still doing the reasoning behind the scenes. What changes is how people access that intelligence.
The real shift is cultural: we’re moving toward a world where people speak to machines the way they speak to each other. That opens up new jobs no one’s named yet and replaces the ones you might have always considered super safe. At the global level, voice will also change who gets access to services, education, and work — not just how we interact with machines.
Your work is dedicated to simplifying cross-lingual communication for remote communities. How will voice technologies change global communication and democratize access to information in the next 5 years?
Voice won’t change how we communicate, but it will remove the need for intermediaries. Instead of hiring interpreters, people will be able to talk directly across 20-30 languages. That applies to business, education, and even talking to an AI agent on the other side of the world.
Voice doesn’t do anything that text can’t, it just does it faster. But “democratization” doesn’t mean “free.” These systems are resource-intensive and won’t be cheap to run. So, yes, access will expand, dramaticall, but primarily for people and companies that can afford to pay.
For everyone else, free services will exist, but they’ll come with tradeoffs. As always, if something in the digital economy is free, then more likely than not, you’re the product.
