Research that shapes tomorrow's AI
Turning Scientific Research into Real-World AI
Innovation in Artificial Intelligence emerges at the intersection of scientific research and industrial application. That is why we collaborate with universities, research centers, and specialized organizations to explore new AI architectures, develop advanced training methodologies, and address the challenges related to the reliability, safety, and governability of intelligent systems.
The expertise and results generated through these collaborations contribute directly to the evolution of our Velvet family of models, our multimodal models, and the AIWave platform.
Foundation Models & Advanced Architectures
Research is a foundational element in the development of our AI technologies.
We collaborate with the academic world to study new Large Language Model architectures, Mixture of Experts (MoE) approaches, innovative learning techniques, and optimization methodologies aimed at improving model performance, efficiency, and sustainability.
These activities contribute to the ongoing evolution of the Velvet family and to building systems that are increasingly adaptable to complex enterprise contexts.
AI Alignment, Governance & Responsible AI
Building intelligent systems requires far more than the ability to generate accurate responses.
We collaborate with universities, research centers, and multidisciplinary experts to study AI Alignment techniques, bias assessment, model robustness, and governance methodologies that enable the development of reliable, controllable systems consistent with the goals and values of the organizations that adopt them. Our research activities also address the ethical, regulatory, and organizational implications of Artificial Intelligence, with particular attention to transparency, accountability, risk management, and the protection of individuals' rights.
In this context, we contribute to the development of approaches and frameworks that promote the conscious and responsible use of AI technologies, in line with the evolution of international standards and regulations. The results of these initiatives contribute directly to the evolution of the Velvet models, multimodal systems, and agentic architectures that power the Almawave Labs ecosystem, supporting the development of AI solutions that are reliable, verifiable, and suited even to the most critical and regulated contexts.
Speech, Vision & Multimodal Intelligence
Research is not limited to language alone.
We take part in initiatives focused on developing technologies for speech understanding, automatic transcription, computer vision, and multimodal analysis, contributing to the development of systems capable of understanding information from text, images, documents, and voice.
These activities enrich our ecosystem of proprietary models and cognitive services, enabling the development of increasingly sophisticated AI applications.
Our Funded Projects
Our Funded Projects represent an additional way through which we turn scientific research into concrete innovation. Through national and European initiatives, we collaborate with universities, research centers, institutions, and industrial partners to develop and test new Artificial Intelligence technologies, accelerating their transfer into real-world applications and products.