Will voice AI assistants succeed? Seven trends predicted by serial entrepreneurs in the consumer AI market

Emotional AI application founder Eugenia Kuyda recently proposed a series of predictions for the future of “Consumer AI (Consumer AI),” pointing out that many current popular directions may be difficult to succeed and highlighting the key trends with true scalability potential. These include perspectives such as “Voice and lifestyle recording AI devices will not succeed,” providing industry with valuable insights.

Overcoming tragedy to become a continuous AI entrepreneur: Who is Eugenia Kuyda?

Eugenia Kuyda is a Russian-born entrepreneur and artificial intelligence innovator, best known for founding the AI emotional companionship app Replika, which has accumulated over 40 million users. Since this app was originally created to commemorate a deceased friend, it is not surprising that he approaches the consumer market from the perspective of “AI as a friend.”

Source: TED Talks

Kuyda also founded Wabi in 2025, a social platform promoting the creation and sharing of mini-apps (Mini-apps) that anyone can generate. Unlike the market’s common focus on efficiency and productivity tools, she has long explored the relationship building and emotional connection between humans and AI, starting from psychological aspects, and offers more pragmatic views on the success factors of consumer AI.

Seven Major Predictions

Voice AI devices will struggle to replace smartphones

Firstly, regarding the recent rise of screenless AI devices, such as voice-oriented wearable devices, Kuyda bluntly states that survival for these products will be very difficult.

She points out that most people’s core behavior when using smartphones is passive information reception, such as scrolling through content and messages, and voice interaction is not suitable as the main interface: “Users tend to speak only when alone, which increases the cost of initiating each interaction.”

She admits that, given the high integration of smartphone functions and current high user engagement, the market finds it hard to justify “carrying another device.”

“Real-time listening devices” will also not succeed

At the same time, Kuyda believes that AI devices that record and monitor daily life around the clock will also fail: “Most daily conversations are not important enough to be constantly recorded, and truly important situations like meetings or dates are recorded by users themselves on their phones.”

In contrast, data such as emails, subscription records, and consumer behavior often better reflect a person’s true state and needs. She also notes that if AI mobile devices shift to full video recording, social acceptance and privacy boundaries will be another challenge.

(Foxconn replaces Lianxun to win OpenAI AI mobile device order, latest details reveal “mini camera”?)

Mini-apps unlock the potential of “user-generated software”

In her predictions, the biggest change comes from the transformation of software production methods. Kuyda believes that mini-apps will lead software development toward a path similar to short video content creation:

Traditional app development involves login systems, databases, and payment integrations, which are complex and have high barriers. However, most people can clearly describe “needs, processes, and screens.”

When the cost of creating personalized tools is lower than downloading and learning new apps, user-generated “personal software (UGC software)” will enable a variety of new consumer applications to flourish.

(Robinhood CEO: AI will make “one-person companies” the norm, and the wave of brand tokenization will rise)

By 2030, there will be two major general AI chatbots

Kuyda predicts that future general-purpose AI will not be dominated by a single product, but will have two distinctly different roles:

The first is similar to ChatGPT-like tool assistants, focusing on search and task handling; the second is centered on companionship and guidance, as “AI friends” with emotional connections and a certain degree of unpredictability.

She believes these two types of AI are conflicting in product design, making it difficult to integrate them into the same system.

(Artificial Intelligence Research: About 30% of American teenagers use AI chatbots daily, safety concerns increasing)

App advertising models will become ineffective

On the business model front, Kuyda also warns about the advertising-centric app business model. She points out that many studios rapidly copy popular apps to flood the market with ads, converting user data before users have truly experienced the product.

This approach compresses the importance of the product itself, ultimately eroding the arbitrage space between “cost to acquire a user (CAC)” and “total revenue contributed by that user (LTV),” making it unsustainable for long-term business.

“AI video companionship” may be the fastest-growing consumer AI product

From a commercial perspective, Kuyda expresses optimism about AI video companionship products: “Once real-time image generation costs decrease, combined with unlimited personalization and 24/7 interaction capabilities, AI will meet the highly scalable companionship demand, ultimately creating consumer products with annual recurring revenue (ARR) exceeding $1 billion.”

Looking back, relying solely on real human scale like OnlyFans will seem insignificant.

Who can tell the public “what AI can do” will be the winner

Finally, she points out that the biggest bottleneck for AI today is not capability but “discovery (Discovery).” For most users, a blank input box still resembles a command line interface (Command Line), suitable only for chatting and searching; however, many potential applications are hidden behind prompts:

Whoever can figure out how to unlock all these application scenarios and demonstrate what artificial intelligence can do to the public will become the next major consumer platform.

Technology rooted in humanity: Consumer AI competition returns to understanding human nature

Overall, Kuyda’s seven predictions are based on human behavior and psychology rather than technology. From hardware forms and interaction interfaces to business models, she repeatedly emphasizes a core premise: successful consumer AI must align with human nature and create demand accordingly.

In the rapidly evolving industry, these calm observations may also help investors identify the next winners.

This article “Will Voice AI Assistants Fail? Seven Trends in the Consumer AI Market Predicted by a Serial Entrepreneur” first appeared on Chain News ABMedia.