IntuiCell, a Nordic deep-tech startup at the forefront of blending neuroscience and AI innovation, is announcing a world-first scientific and technological milestone: the development of digital intelligence capable of real-world, real-time learning.
This breakthrough by Intuicell, a spin-out of Sweden’s Lund University, marks a paradigm shift in its field: moving beyond static machine learning models (the mainstay of traditional AI) to a fully functional “digital nervous system” capable of scaling naturally to human-level intelligence.
Earlier this month, IntuiCell released the first video footage of an off-the-shelf robot quadruped (with no pre-programmed intelligence or instructions) which the company has augmented with a digital nervous system.
In doing so, IntuiCell has created, for the first time in history, a physical AI agent capable of learning autonomously in a manner which mimics a biological nervous system.
Unlike traditional AI models that are bound by static training data, the robot dog – dubbed “Luna” – perceives, processes, and improves itself through direct interaction with its world.
IntuiCell’s video of Luna demonstrates the first functional digital nervous system in action: viewers witnessing how she learns to control her body and teaches herself to stand, like a newborn animal, through trial, error, and cumulative experience.
Since the 1950s, the scientific community has made several failed attempts to replicate the mammalian brain into software; the most recent notable attempt being the Brain Mind Institute’s defunct Blue Brain Project.
Now, following 30 years of contrarian neuroscience research at Lund University, IntuiCell has become the first to achieve this by developing its first major component, which replicates the core functionality of the spinal cord.
The Swedish company also has the functioning basis for technology that will operate as the Thalamocortex (the part of the brain which processes and predicts the world around us).
This breakthrough will enable the first real-world teachable systems; machines that learn from us, in the same way as we would teach a new skill to an animal.
To further demonstrate this concept in action, IntuiCell will be the first AI company in the world that will hire a dog trainer – rather than integrating an LLM and vast datasets – to teach a physical agent new skills.
Surpassing previous efforts towards realising the original aspirations for truly intelligent machines, IntuiCell expects to complete the full digital nervous system within the next two years.
Ultimately, the software platform will be able to equip any agent, physical or digital, with lifelong learning and adaptation to the unknown – capabilities once considered unique to biological creatures.
Rendering traditional AI ‘obsolete’
Currently, practically all AI models rely on backpropagation (the training process of feeding error rates back through a neural network) and predefined datasets, limiting their ability to learn dynamically.
From LLMs, to deep reinforcement learning, as well as some outliers that claim brain-inspired approaches, the entire field of AI is dominated by foundation models trained on vast datasets.
Whilst these approaches have achieved remarkable results in static, low-stake applications (such as generating text and images), all have remained incapable of replicating biological intelligence’s ability to solve unseen problems or engage in high-stakes applications (such as robotics or industrial automation).
IntuiCell’s breakthrough promises to eliminate these constraints by using fully recurrent networks powered by a decentralised learning algorithm based on the learning patterns of the brain.
This approach enables AI agents to learn from their own experiences (without needing preloaded data or backpropagation), adapt in real time to new environments and challenges, and scale naturally towards human-level intelligence.
Viktor Luthman, CEO and co-founder of IntuiCell, says: “For decades, AI has excelled at processing vast amounts of data but has fundamentally been incapable of real intelligence.
“Our system changes that. We’ve translated the principles of biological learning into software, enabling AI to evolve, adapt, and interact with the world in ways never previously envisaged.
“Through licensing our digital nervous system, we aim to become the infrastructure for all non-biological intelligence – empowering others to solve real-world problems we cannot foresee today, without a reliance on massive training datasets.”
Udaya Rongala, researcher and co-founder of IntuiCell, says: “IntuiCell’s AI is not just an improved version of machine learning; it is an entirely new category of intelligence.
“Our work is rooted in 30 years of contrarian neuroscience research and built on a unique understanding of how intelligence emerges from the architecture and dynamics of the nervous system as a whole.
“The obsession with brute-force scaling, billions of parameters, more compute, and more data is an artifact of a fundamentally wrong approach to achieving intelligence. IntuiCell is not chasing a bigger-is-better paradigm. Intelligence is not our end-goal, but our starting point.”