How to Build Emotionally Intelligent AI: Insights from Marie Toft of Emotionise

What does it mean to build AI that understands human emotion? In a world increasingly shaped by generative AI, Marie Toft, CEO and co-founder of Emotionise, argues that emotional intelligence is not merely a "soft skill"—it is a power skill that will define the future of work.

In this episode of Chatting GPT, host Maryrose Lyons sits down with Marie to explore how her company is training AI to be emotionally intelligent, and why this matters for businesses, customer experience, and the very nature of human communication.

From Psychotherapy to AI: An Unlikely Journey

Marie Toft brings a unique perspective to the AI industry. With a background as a psychotherapist and veteran television content creator, she has spent decades understanding how humans communicate and connect. This expertise became the foundation for Emotionise, a company on a mission to bring kindness, compassion, and humanity to content and communications through groundbreaking AI.

"Emotional intelligence, for years, was seen as a kind of soft skill," Marie explains. "Quite often it's seen as a female skill, which I think is really misplaced. I would argue emotional intelligence is a power skill."

Her insight is particularly relevant as AI becomes more embedded in our professional lives. The ability to communicate with empathy and understanding—traits that machines have historically struggled to replicate—may become the differentiating factor that makes AI truly useful in human contexts.

Building EMUR: The Emotionally Intelligent Model

At the heart of Emotionise's work is EMUR, an emotionally intelligent model that acts as a cognitive collaborator for better communication. Unlike generic large language models, EMUR was developed specifically to understand and enhance emotional intelligence in written communications.

The development process was rigorous. Emotionise partnered with Open AI from 2021, working with GPT-3 to create a model trained on proprietary data. But this was not a simple case of feeding the model vast quantities of text. Marie and her team took a quality-over-quantity approach, working with writers who had passed rigorous emotional intelligence testing to create training data that genuinely embodied empathetic communication.

The result is a tool that integrates directly into everyday workflows, functioning as a live cognitive collaborator inside Microsoft Outlook and Gmail. When a user drafts an email, EMUR analyses the text and suggests rewrites that make the communication more empathetic—without changing the factual content or core message.

The Business Case for Emotionally Intelligent AI

Marie makes a compelling business case for emotionally intelligent AI. Research shows that only 36% of people naturally exhibit high emotional intelligence. For the remaining 64%, AI-assisted tools like EMUR offer genuine value in improving workplace communication.

The impact extends beyond internal communications. Companies using emotionally intelligent AI in customer-facing roles have seen measurable improvements in Net Promoter Scores (NPS) and overall customer satisfaction. When customers feel understood and valued—when the communication they receive demonstrates genuine empathy—their relationship with the brand strengthens.

"As AI becomes more and more part of our lives, that level of humanity and emotional intelligence is going to become even more important," Marie notes. This insight has profound implications for organisations navigating AI adoption.

Regulation, Ethics, and the EU AI Act

The conversation also touches on the regulatory landscape, particularly the EU AI Act. As governments worldwide grapple with how to govern AI systems, emotionally intelligent AI sits at an interesting intersection. The technology deals with human emotions—a sensitive domain that requires careful ethical consideration.

Marie's background in psychotherapy informs Emotionise's approach to these challenges. The company is building AI that augments human capability rather than replacing human judgment. EMUR suggests; it does not dictate. The human user remains in control, making the final decision about what to communicate and how.

This human-in-the-loop approach aligns with emerging best practices for responsible AI deployment and positions Emotionise well as regulatory frameworks mature.

Looking Forward: The Future of Human-AI Collaboration

What emerges from this conversation is a vision of AI that enhances rather than diminishes human connection. Marie Toft's work demonstrates that it is possible to build AI systems that understand emotional nuance—systems that can help us communicate more effectively while preserving the authenticity of human expression.

For organisations exploring AI adoption workshops and AI strategy for leadership teams, emotionally intelligent AI represents a practical and valuable application. It addresses a genuine business need—better communication—while respecting the complexity of human interaction.

As we continue to integrate AI into our workplaces and workflows, the question is no longer whether machines can be intelligent. The question is whether they can be emotionally intelligent—and Marie Toft's work suggests that the answer is yes.

Want the full conversation? Watch the Chatting GPT episode on YouTube here: https://www.youtube.com/watch?v=T6ZVM_kSyy8