Artem Rodichev is the Founder and CEO of Ex-human, an organization targeted on constructing empathetic AI characters for participating conversations. Earlier than founding Ex-human, Artem was the Head of AI at Replika from 2017 to 2021, the place he led the event one of the common English-speaking chatbots, rising its person base to 10 million within the U.S.
You spent a number of years as Head of AI at Replika, constructing one of the common conversational AIs. What impressed you to go away and located Ex-human, and the way did your expertise at Replika affect your imaginative and prescient for Ex-human?
Throughout my time at Replika, I had the chance to assist form a conversational AI that resonated with hundreds of thousands of customers, which gave me deep perception into how folks join with expertise on an emotional stage. As rewarding because it was, I at all times felt there was extra potential for AI past what we had been constructing. I noticed a possibility to broaden AI’s potential throughout a number of sectors and use circumstances.
At Replika, the main target was totally on making a companion AI, however I started to check how conversational AI may very well be built-in into numerous purposes, providing personalised, emotionally clever interactions for a wide selection of use circumstances — from leisure to enterprise. That realization sparked the thought for Ex-Human, the place we goal to develop AI expertise that not solely gives companionship but in addition powers modern experiences for companies and shoppers alike.
Ex-Human was born from the will to push the boundaries of AI even additional, making it extra adaptive, participating, and able to reworking how folks work together with digital characters throughout numerous industries.
Ex-human makes use of AI avatars to interact hundreds of thousands of customers. Are you able to clarify how your AI avatars are designed to be empathetic and responsive?
We’ve spent loads of time refining how our AI characters work together with customers, making certain that they’re not simply reactive however really attentive to the emotional tone of the dialog. A technique we do that is by coaching our fashions on hundreds of thousands of open-domain conversations, permitting them to discover ways to interact in a manner that feels pure and emotionally conscious.
The aim is to make customers really feel understood and heard. These characters aren’t simply programmed to offer solutions; they’re designed to foster genuine interactions. They decide up on delicate emotional cues and reply in ways in which really feel human-like and comforting. That’s why our customers spend vital time interacting with them.
Past simply textual content, our avatars are able to multi-modal interactions. They’ll reply with voice, ship contextual photos, and even video, which provides a layer of richness and depth to the conversations. This number of interplay kinds helps customers really feel extra immersed within the expertise and permits them to construct full tales with the AI characters. On the finish of the day, we goal to create AI that goes past customary interactions, providing customers a deeply participating, emotionally clever expertise that retains them returning.
How does the info collected out of your B2C platform botify.ai affect the coaching and improvement of your AI fashions?
With hundreds of thousands of customers participating in conversations with all kinds of characters on our platform, we’re in a position to accumulate an infinite quantity of real-world interplay information. This helps us perceive not solely how customers talk, but in addition what makes these interactions emotionally significant.
By analyzing this information, we’re in a position to fine-tune our fashions to be extra responsive and empathetic. It’s a steady cycle: person interactions present insights, permitting us to fine-tune and evolve the AI. This method permits us to enhance the conversational movement, tone, and the general emotional intelligence of our avatars.
In essence, our B2C platform is a sort of real-time testing floor. The info we accumulate permits us to continuously evolve and adapt our AI to fulfill person expectations, and people enhancements instantly affect how we develop our B2B options as effectively. It’s how we be certain that our AI stays not solely cutting-edge but in addition deeply human in its interactions.
How does Ex-human guarantee information privateness and moral AI practices whereas utilizing massive quantities of person information for coaching functions?
Knowledge privateness and moral AI practices are our prime priorities, particularly given the quantities of person information we work with. Firstly, we be certain that all information is anonymized earlier than it’s used for any coaching or evaluation. This implies private identifiers are eliminated, so the info can’t be traced again to particular person customers, defending their privateness all through the method.
We even have strict information dealing with protocols in place. Customers’ information isn’t used with out their consent, and we’re absolutely clear about how the info is collected and utilized. Moreover, we often evaluate our practices to remain in step with evolving information safety legal guidelines and moral pointers.
In the case of coaching our AI, we place a powerful emphasis on making certain that our fashions behave in a accountable and unbiased manner. By continuously monitoring and testing our AI, we work to stop any unintended biases from showing in our interactions. This mix of privateness safeguards and moral AI improvement ensures that we are able to ship emotionally clever and responsive AI with out compromising person belief or safety.
Ex-human’s avatars have considerably greater person engagement than platforms like Instagram. What’s the important thing to attaining such excessive ranges of person interplay?
The important thing lies within the emotional responsiveness – it’s what retains customers coming again and spending extra time interacting with our avatars.
One other issue is the extent of customization and personalization we provide. Customers can create and work together with all kinds of AI characters, every with its personal distinctive character and magnificence. Whether or not it’s chatting with a fictional character or creating a very personalized avatar, customers really feel extra concerned and related to the expertise, making it way more participating than merely scrolling by means of content material on social media platforms.
Moreover, our give attention to multimodal interactions, the place avatars reply with textual content, photos, audio, and even video, provides to the immersive expertise. This makes customers really feel actively concerned in a story that adapts and grows with them, reasonably than passively consuming content material. All of those parts collectively contribute to the upper ranges of engagement we see in comparison with social media platforms.
You’ve fashioned partnerships with firms like Grindr. How do companies profit from integrating Ex-human’s AI, and what measurable outcomes have they seen?
We permit companies to supply extra personalised and fascinating person experiences. Our expertise permits them to construct characters that interact customers in considerate, pure conversations. For instance, within the case of Grindr, we’ve launched options like an AI wingman that helps customers enhance their profiles and a digital companion to observe relationship conversations. These options not solely make the platform extra participating but in addition present sensible worth to customers.
By way of measurable outcomes, firms have seen enhancements in person engagement and retention after integrating our AI options. This reveals how AI-driven characters can add actual worth to companies by creating deeper connections with customers.
How does your AI-driven engagement examine to conventional advertising and marketing strategies by way of buyer retention and conversion?
We’re in a position to provide one thing way more immersive and personalised. Conventional advertising and marketing typically focuses on short-term consideration grabs or delivering static messages, however our platform is constructed round steady experiences—which permits for deeper engagement over time.
What actually units us aside is the extent of interplay. On common, customers spend over 60 minutes a day with our avatars, and our paying customers interact for even longer, exchanging a whole bunch of messages. The mix of multimodal capabilities creates a way more participating expertise, akin to an AI-powered Netflix, the place customers grow to be a part of the story. This goes past what conventional advertising and marketing can obtain.
How do you see the way forward for AI avatars and digital people reworking industries like customer support, healthcare, or leisure?
In customer support, they might deal with routine queries extra effectively whereas sustaining a excessive stage of consistency, liberating up human brokers to give attention to extra advanced points. This mix of automation and personalised assist can considerably enhance response instances and total buyer satisfaction.
In healthcare, AI avatars might help with psychological well being assist, providing companionship and even teaching, significantly in eventualities the place human professionals are stretched skinny. They is also used to information sufferers by means of administrative processes, making healthcare extra accessible and fewer intimidating for individuals who would possibly in any other case keep away from searching for assist.
In leisure, the chances are even broader. AI avatars can create solely new kinds of interactive content material, permitting customers to interact instantly with characters in ways in which conventional media can’t provide. Whether or not it’s a recreation, a digital live performance, and even digital influencers, AI can generate distinctive experiences which can be tailor-made to particular person preferences, opening up new income streams for the trade.
What key classes have you ever realized out of your transition from Head of AI at Replika to founding and scaling your individual firm?
A key lesson I’ve realized within the shift is how dramatically the position adjustments. As an engineer, I might dive deep into technical challenges and give attention to problem-solving, however as a founder, the scope expands considerably. All of the sudden, you’re not simply constructing expertise—you’re additionally managing groups, working with buyers, making strategic choices, and making certain the corporate’s development on a number of fronts.
The hardest half was shifting from a hands-on, technical mindset to 1 that includes continuously switching between totally different roles and tasks. It’s much less about fixing one particular drawback and extra about managing the whole lot concurrently, all whereas retaining the general imaginative and prescient in sight. This broader perspective has been important for rising Ex-Human and navigating the entrepreneurial panorama.
Thanks for the nice interview, readers who want to be taught extra ought to go to Ex-human.