TYRIS AI SL

Their core work

TYRIS AI is a Valencia-based AI software company specialising in explainable and graph-based machine learning for industrial applications. Their core work involves building decision support systems that make AI reasoning transparent and auditable — a critical requirement in regulated or safety-sensitive manufacturing environments. In XMANAI they contributed explainability and graph deep learning components for manufacturing AI pipelines; in TEAMING.AI they worked on frameworks that allow humans and AI systems to collaborate and co-evolve over time. Their practical value lies in turning opaque machine learning models into tools that production engineers can actually trust, interrogate, and maintain.

What they specialise in

How they've shifted over time

Both H2020 projects started within a single year (2020–2021), so a longitudinal shift within the H2020 programme is not directly observable — the dataset is too compressed for a clear before/after story. That said, there is a detectable thematic progression: their first project (XMANAI) is centred on model-level techniques — explainability, graph learning, hybrid ML — while the second (TEAMING.AI) moves up the stack toward system-level concerns: how humans maintain, oversee, and evolve deployed AI. This suggests TYRIS AI is maturing from building AI components toward addressing the full lifecycle of AI systems in industrial use.

TYRIS AI appears to be moving toward AI governance and lifecycle management in manufacturing — a space with growing regulatory demand under the EU AI Act — which makes them a relevant partner for any consortium needing trustworthy, auditable industrial AI.

How they like to work

TYRIS AI has participated exclusively as a consortium partner, never taking the coordinator role, which is typical for a focused technical SME that contributes specialist components rather than leading project management. Their two projects sit within large RIA consortia — XMANAI and TEAMING.AI each involve multiple academic and industrial partners — suggesting they are comfortable operating within complex, multi-stakeholder research environments. Working with them likely means engaging a compact, technically sharp team that delivers specific AI modules within a broader system architecture.

TYRIS AI has built connections with 33 unique consortium partners across 11 countries through just two projects, indicating both projects were mid-to-large consortia with broad European participation. Their geographic footprint is fully European, with no evidence of a national bias beyond their Spanish base.

What sets them apart

TYRIS AI occupies a narrow but increasingly valuable niche: explainability and graph-based reasoning for industrial AI, not just AI in general. While many SMEs offer generic machine learning services, TYRIS AI's dual focus on making models interpretable and on human-AI co-evolution positions them directly at the intersection of the EU AI Act's transparency requirements and the practical needs of manufacturers adopting AI. For a consortium building a responsible AI application in Industry 4.0, they offer technical credibility that a general-purpose AI consultancy cannot match.

Highlights from their portfolio

• XMANAI
  Their highest-funded project and the one that most clearly defines their technical identity — combining explainable AI, graph deep learning, and hybrid ML specifically for manufacturing decision support, a technically ambitious combination.
• TEAMING.AI
  Addresses the harder long-term problem of how AI systems evolve alongside human operators, signalling TYRIS AI's ambition beyond model-building toward AI governance and maintenance platforms.