Interhuman AI, a generative artificial intelligence company researching how to make AI models more socially intelligent, today announced it has raised €2 million, about $2.3 million, in a pre-seed funding round led by PSV Tech.

EIFO, Antler, Yope, and angels from Ada Ventures also participated in the funding round for the AI social intelligence company.

When humans speak to one another, not all of their communication is done through the words — that is to say, much is unsaid. In any given conversation, a person might fidget, smile, shrug their shoulders, glance away when discussing an uncomfortable topic or become more expressive with their hands when they’re excited.

Current AI models can be blind to this kind of behavior, which comes naturally to some people. AI conversations can mimic understanding this by taking cues from words, which express the basic level of meaning in a conversation, but they miss the subtext and signals that come from the person’s voice and behavior.

To manage this, Interhuman is adding that layer using AI vision and audio capabilities to help understand body language and expression during conversation in real-time. This will allow specially trained models to combine signals from facial expressions, arm and hand movements, voice inflection and tonality, and more into “social signals.”

“The global AI market is growing fast. However, to realize the full potential of AI solutions, it needs to take into consideration all aspects of human behaviour,” said co-founder and Chief Executive Paula Petcu. “AI isn’t just about efficiency and workflow optimisation — it is about human interaction.”

Interhuman said the investment will allow the company to hire more engineers to fuel the development of its frontier models and services, which it’s deploying to enhance AI applications’ understanding of human communication.

“We’re only just beginning to see what happens when social intelligence is built into AI,” said Alexander Viterbo‑Horten, partner at PSV Tech. “This breakthrough has the potential to transform healthcare, education, and any sector where human‑AI interaction matters.”

The company said it intends to start with use cases where “the stakes are the highest,” training and coaching, healthcare and customer interaction. All of these roles, AI has high human-interaction and customer-facing roles. In many of these cases, users find themselves interacting with AI that must quickly determine whether its communication is understood.

For instance, an AI-driven tutoring app could identify when a student is confused and adjust the session pace to match their learning speed. This would reduce the robotic sense of urgency that AI apps tend to create and allow users to go at their own pace without having to demand the AI slow down for them.

In a healthcare or customer service role, the AI could tell when a patient or customer is becoming frustrated or unable to follow along. In many of these apps, AI agents are only the front line in helping with tasks and explaining basic information. When things become more complex, it’s their job to bring on a human who can provide the personal touch. Increasing frustration or discomfort might not be readily noticeable in polite words, but it might be revealed quickly in a tight smile, hesitation or pursed lips.

As more AI gets deployed into roles where it interacts with people, Interhuman argues, it’s necessary that companies attend more to how their systems treat the people they interact with. Generative AI solutions may feel amazing in comparison to “Press 1 for appointment times,” but they could be a bit more “human.”

Image: SiliconANGLE/Microsoft Designer