Indeep AI is a Spanish research SME, pioneering applied neuroscience and mind modelling to keep the human in the loop of AI decision-making. We join European research consortia as a technical partner. Three EIC-funded projects since 2022, €4.26M awarded.
At the intersection of neuroscience and trustworthy AI, we bridge the gap between machines and human cognition. By applying advanced pragmatic analysis, we decode the hidden communicative intent behind text, while our Theory of Mind models use sensor analytics to infer mental states, adapting AI communication in real-time to ensure maximum comprehension.m of research.
Language is rarely just literal; it is a complex web of implications, hedges, and structural signals. At Indeep AI, our pragmatic analysis research moves beyond vocabulary to uncover the true communicative intent of both human and machine authors.
By applying a formal, Gricean framework to text, evaluating information in terms of quantity, quality, relevance, and manner, we bridge the gap between complex document structures and human understanding. Crucially, we also analyse text to detect epistemic traps that may induce cognitive biases in the user. This enables us to transform static text into transparent, cooperative communication, ensuring that every document is not just read, but fully and accurately understood.
We are building artificial agents that read your mind. The Theory of Mind layer utilises advanced sensor analytics, incorporating Lexical and Pragmatic Analysis, Facial Expressions, Speech Emotion Recognition, Vocal Biomarkers, and Vocal Stress Analysis, to decode your attention, confusion, understanding, and interest in the topic, while also identifying priming effects and cognitive biases.
By treating each interaction as a dynamic cognitive exchange, our agents adapt their communication strategy in real-time, simplifying complex concepts or adjusting the pace of delivery to keep you firmly in control of the decision-making process.
Expert software development and systems engineering across SaaS, high-performance networks, and GPU server infrastructure. We offer deep technical proficiency in LLMs, UI/UX design, databases, and backend processes, underpinned by our core expertise in neural networks and computational linguistics and pragmatic analysis.
With our dual presence in Barcelona and Las Palmas de Gran Canaria, we offer a strategic geographic advantage. Our base in an EU Outermost Region aligns with European policy priorities on territorial cohesion and regional development, a key metric for proposal evaluators, significantly strengthening the competitive profile and funding prospects of any research consortium.
We can carry dissemination, communication and exploitation; we have project management in our DNA.
We are used to carrying substantive shares of project work. We build long-standing relationships with recurring collaborators.
A private participant strengthens the impact and exploitation dimensions that evaluators score, and brings an industrial perspective to a consortium of research organisations.
We are interested in consortia forming for Horizon Europe calls where computational linguistics, cognitive modelling or human-centred AI has a role.
Indeep AI holds the grant as sole beneficiary, running the full cycle of the project, including technical work, field validation, compliance, IP, reporting, and communication.
Indeep AI contributed the ASAC architecture, a cognitive attention control embedded in transformers via a VQ-VAE module, with measured gains to adversarial input.
Indeep AI provides a technical contribution by creating a digital vascularisation map, as well as business development activities by leading the stakeholder engagement initiatives.
Chief Scientific Advisor
Professor of Neuroscience with a focus on Action and Cognition at the University Medical Center Hamburg-Eppendorf
GPT-4o reads the mind in the eyes.
Computers in Human Behavior
Testing theory of mind in large language models and humans.
Nature Human Behaviour
We are interested in consortia forming for Horizon Europe calls where computational linguistics, cognitive modelling or human-centred AI has a role. Early conversations are more useful than late ones, so if a call is on your radar, we would rather hear about it while the consortium is still taking shape.