Launching AI Avatars: The Future of Conversational AI in Education and Product Innovation RealAvatar recently launched Andrew Ng's AI avatar in collaboration with DeepLearning.AI, enabling learners to have a conversation with an AI version of the renowned AI educator for career advice through text chat and voice calls. Dmitry Pyanov, who leads product at RealAvatar, discusses the future of AI in education and the evolving role of product management.

RealAvatar uses conversational AI to capture the style, voice, and expertise of leading educators. In Andrew Ng's case, we wanted to give DeepLearning.AI's (the company was founded in 2017 by Andrew Ng; it empowers over 7 million learners worldwide with AI and machine learning education) global community the chance to interact with him in real time, get personalized career advice and mentorship, and receive support on any educational path.

As Head of Product, Dmitry Pyanov wears many different hats—from crafting RealAvatar's product vision, brand and design to managing engineering ambiguity and prioritizing features that best serve learners. He works closely with a cross-functional team, continually improving conversational quality and building the most authentic representation of Andrew Ng's character.

The outstanding team with a wide range of expertise allows Dmitry to quickly innovate and execute at the highest level. The RealAvatar team brings together deep expertise in AI/ML, consumer technology, and education. CEO Jeff Daniel previously founded consumer app platform StarMaker, publishing the official apps for The Voice and American Idol. The team also includes board member Tim Westergren, founder and CEO of Pandora. Nedelina Teneva, PhD, whose experience spans machine learning leadership at Amazon Alexa and teaching at UC Berkeley, leads the AI team. In 2024, RealAvatar was selected for the highly competitive AWS Generative AI Accelerator, which accepted less than 2% of over 4,700 applicants worldwide.

"In product development, the most important factor is stepping away from internal assumptions and truly understanding what users would benefit from. It's essential to directly learn about their experiences, and see the world from their perspective. This allows you to find the balance between intuition and data-driven insights," — says Pyanov.

"You really see the magic when people start chatting with these avatars. It's not just about dissecting complex topics and streamlining the educational process—it's about empowering learners with personal advice and building a connection that leads to better learning outcomes. Of course, it's not the real Andrew, but in its AI form, it's surprisingly powerful."

Before RealAvatar, Dmitry played a pivotal role at a pioneering conversational AI app Replika, which was the first big AI companion that entered the market back in 2017, attracted over 30 million users and gained global media recognition.

Replika was at the forefront of conversational AI even before the large language model revolution that arrived with the ChatGPT breakthrough. It effectively passed the Turing test in a pre-LLM era by leveraging the existing technologies of the time. The team tested features like voice calls, journaling, avatar customization, and image recognition early on—long before anyone else. Rapid iteration also helped pinpoint what mattered to users, what was feasible, and what the pitfalls could be.

"Before my career in product management, I considered myself a multidisciplinary creative—I was always driven by culture, fascinated with the intersection of design and technology. My first big tech product and marketing role was at Yandex.Music back in 2014, where I worked on a large-scale redesign of the country's most popular music streaming service. Our team introduced social features like public user profiles, user-generated content curation, and playlist sharing, while also overhauling the service's overall interface. It was my first experience at a major tech company with a rigorous engineering culture and very high standards. I learned a lot there."- says Dmitry.

He then worked at several startups before joining the Replika team to lead vision, branding, community, and eventually product management with focus on growth.

"In smaller teams, being a product manager often means wearing many hats, which I really enjoy. You get to take an idea from zero to one—rapidly prototyping, launching, and then iterating based on user feedback, until you have something millions of people use and love." After his work at Replika Pyanov also contributed to one of the most fast growing consumer products — The Pattern, a leading astrology app that reached over 15 million users up to date.

"There are different flavors of product management", he says. "AI product management, especially for conversational or multimodal applications, is fascinating because one is working with a different medium than traditional user interface and experience. It's a new medium of interaction, you get to work on character design, safety, engagement and every day there are new breakthrough technologies that could be productized in non-trivial innovative ways. It's always a challenge to assess feasibility when some new technology becomes available and there's lots of hype around it, sometimes prematurely, sometimes misleading in terms of real market adoption."

The rise of AI tools has changed the entire product management landscape. Things like data-science, research, design, and coding have become widely accessible. As of 2025 as a product manager you can prototype things of great complexity within hours. Still, a product manager's main value remains being creative, deciding what to build and why, balancing technical feasibility with user value, and setting clear priorities for the team.

Product manager has to identify the most important metrics, define a viable scope, and iterate. Data-driven decisions alone aren't enough; you also need strong product instincts and empathy for users, plus an awareness that your own opinions may not reflect the much wider set of use cases out there. "When working on zero-to-one moonshots with smaller agile teams, you have to rely on your instincts, try, fail and repeat, and execute rapidly, especially in this era of rapid innovation driven by the LLMs arms race changing the rules and established culture in tech.", — Dmitry concludes.