ALGEBRAIC AI SL

Their core work

Algebraic AI is a Madrid-based private company specializing in mathematically rigorous, human-centric approaches to artificial intelligence and machine learning. Their core work revolves around algebraic machine learning — a research-intensive method that applies abstract algebra to build AI systems that are interpretable, auditable, and aligned with human values rather than opaque neural black boxes. They contribute to large European research consortia as a technical specialist, bringing formal mathematical methods to problems of AI ethics, transparency, and cognitive alignment. Their work sits at the intersection of theoretical computer science and applied AI, targeting both foundational research and real-world deployment concerns.

What they specialise in

How they've shifted over time

Both of Algebraic AI's projects began in 2020, so their H2020 trajectory is short but reveals a meaningful sharpening of focus. Early project keywords — human-centric AI, ethical AI, ubiquitous computing — reflect broad participation in the European discourse on responsible AI through HumanE-AI-Net, a large network consortium. Their more recent keyword, human-centric algebraic machine learning, signals a move from general ethical AI advocacy toward a specific, proprietary technical method: applying algebraic structures to make ML systems formally interpretable. The trend is from broad positioning in the AI-ethics conversation toward owning a distinct technical niche in formal, mathematically grounded ML.

Algebraic AI is consolidating around a specific formal-methods approach to interpretable ML, positioning itself as a technical specialist rather than a generalist AI ethics voice — making them an increasingly targeted fit for consortia needing rigorous, theory-grounded AI contributions.

How they like to work

Algebraic AI has participated in both projects as a consortium member, never as coordinator, which is consistent with a specialist contributor role — they bring a focused technical capability rather than managing large partnerships. Their involvement in HumanE-AI-Net, one of Europe's largest AI research networks, exposed them to an unusually wide partner base: 63 unique collaborators across 21 countries from just two projects. This suggests they are comfortable operating within large, diverse consortia and are likely valued for a specific technical contribution rather than administrative leadership.

Despite only two projects, Algebraic AI has built a notably broad network of 63 unique partners spanning 21 countries, driven largely by HumanE-AI-Net's pan-European scope. Their reach is genuinely European, with no apparent geographic concentration beyond their Spanish base.

What sets them apart

Algebraic AI occupies a rare niche: a private company applying abstract algebra and formal mathematics directly to machine learning — a field dominated by statistical and neural approaches. While most AI companies in H2020 work on applications or ethics policies, Algebraic AI works on the mathematical foundations of interpretability, which is increasingly valuable as EU AI regulation demands auditability. For a consortium that needs credible, theory-grounded AI work rather than another neural network benchmark, they offer a differentiated technical identity that few private companies in Europe can match.

Highlights from their portfolio

• ALMA
  Their largest project by far (€717,500, running to 2025), ALMA is the clearest expression of their core technical identity — algebraic machine learning — and likely where their most original IP is being developed.
• HumanE-AI-Net
  Participation in this flagship FET-funded European AI network gave a small private company access to one of the continent's widest AI research consortia, building a partner network disproportionate to their size.