A face for the conversation
For years, most business AI has been text in a chat window or a voice on the phone. Live AI avatars add a third channel: a lifelike digital person who joins a video conversation, listens, and answers in real time with a natural voice and matching facial expression.
Analysts now treat this as a category in its own right. Gartner published its first Market Guide for Digital Humans in October 2025, advising CIOs and innovation leaders to consider digital humans to improve engagement and streamline operations, describing them as AI avatars that deliver humanlike interactions and consistent, scalable service. Gartner's review pages for the market list conversational AI, speech recognition, synthetic voice, multimodal language models and animation that synchronises gestures with speech among the core features.
Customer service leads the way
Customer service is where the case is clearest. In April 2026 Gartner published research on using digital humans to boost customer engagement, noting that service teams struggle to offer support that is both scalable and empathetic, and that AI avatars can provide personalised, 24/7, multimodal experiences while also improving the experience of human agents. An earlier Gartner paper, from November 2025, argued that avatars could free customer service representatives to focus on complex, high-value work.
Businesses appear to agree on where to start. A 2025 German study based on 289 conversations with companies found customer service was the most-cited use for AI avatars (75 per cent), ahead of marketing and sales (62 per cent) and training (43 per cent).
Sales and training: early, but growing
Training is a natural fit because role play is costly to organise. A 2025 study in the Muma Business Review noted that traditional sales role play needs two people to coordinate diaries, and often a third to observe. The same study also offers a reality check: only 40 per cent of invited students used the AI avatar training on offer, although those who did showed a statistically significant link between session length and performance at their school level.
Wider adoption is still at an early stage. Alexander Group's 2025 Sales Pulse Survey found that only 20 per cent of companies use AI to create sales training content, and 9 per cent to deliver it, with most of those deployments less than a year old.
The caveats worth knowing
Live avatars are not a shortcut to results. Gartner's May 2026 assessment of agentic AI in customer service found that while most organisations are exploring the technology, production deployment and financial returns remain limited. Trust matters too: the 2025 iiMedia Research survey of businesses using digital humans found data security and privacy protection to be the biggest challenge, followed by regulation and ethics.
Regulation is catching up. Article 50 of the EU AI Act sets transparency duties, including that people should be told when they are interacting with an AI system. For anyone deploying an avatar that talks to customers, being open about what it is should be the default, not an afterthought.
Tessa: our live avatar at work
At SentientFlow we build secure, custom AI agents for voice, chat and real-time video, working with clients internationally from the UK and South Africa. Tessa is our own live AI avatar, and the clearest way to show what the technology can do is to use it ourselves.
Tessa joins video calls on all the main meeting platforms. She listens, answers questions from a knowledge base of our own documents rather than guesswork, and speaks in a natural voice with a lifelike face. When the conversation is over and someone says thank you, she steps back politely. Afterwards, she writes up meeting notes, which can be emailed automatically.
Tessa runs on our own server and is managed through a simple web control panel. That matters for the caveats above: grounding answers in approved documents keeps responses accurate, and controlling the set-up helps keep data where it belongs.
The aim is not to replace people in important conversations. First calls often cover the same questions again and again; a live avatar can handle those consistently, so our team can spend their time where human judgement counts.
What this means for your business
The pattern from the research is consistent: start with a clear, repeatable use case, ground the avatar in your own knowledge, be transparent with the people it talks to, and measure the outcome. Customer service enquiries, first-stage sales conversations and training role play are all sensible places to begin.
If you would like to see a live avatar in action, or explore what one could do for your team, get in touch with SentientFlow.
#aiagents #customerexperience #digitalhumans #liveaiavatars #sentientflow