AI education startup Diotima spun off from Trinity University after an 18-month incubation period.

This AI company, born out of Trinity, aims to solve teachers' most pressing challenges.

Author and source: Duojing

On May 25, 2026, Trinity College Dublin in Dublin, Ireland, announced that Diotima, an AI education platform founded by former secondary school teacher Siobhan Ryan, has officially spun out from the university to become an independent company. The project has received €500,000 in commercialization funding from Enterprise Ireland, Ireland’s national enterprise development agency, and has completed the full journey from research concept to product launch with support from Learnovate Centre, Trinity’s education technology research center.

Compared to typical educational AI startup projects, Diotima has several distinctive characteristics: its founders are not technologists but frontline teachers; the product was designed from the outset to strictly comply with EU AI regulations and data governance requirements; and its entire commercialization process has been deeply involved and continuously driven by the university’s industry-academia translation system. The combination of these three elements makes Diotima not just a newly established edtech company, but rather a new archetype of educational AI emerging in Europe.

From left to right: Jonathan Dempsey, Tom Pollock, Siobhan Ryan, and Dr. Ann DeWitt

Siobhan Ryan spent twelve years working for the beverage giant Diageo before switching careers to become a secondary school teacher in her thirties. She quickly realized that providing personalized feedback to every student was impossible in large classrooms. Educational research shows that timely, targeted formative feedback significantly enhances learning outcomes, but a single teacher managing thirty to forty students simply doesn’t have the capacity to tailor feedback for each individual. Personalized feedback that identifies specific issues and offers actionable improvement guidance became a luxury.

Ryan’s Diotima core features include: enabling teachers to quickly create assessments, generate differentiated evaluations, provide formative feedback to students, and support the evaluation of the learning process. A teacher needs to prepare exercises of varying difficulty for students at different levels—Diotima can automatically generate these based on learning objectives and student profiles. A teacher needs to grade assignments individually and write personalized comments—Diotima can analyze student responses and generate targeted feedback recommendations.

The key is that the system preserves the teacher’s authority. Content generated by the platform requires teacher review and adjustment, feedback suggestions must be approved by the teacher, and the final evaluations received by students are still issued by the teacher. AI serves as an auxiliary tool, freeing teachers from repetitive, mechanical tasks so they can focus on teaching activities that require professional judgment and emotional connection.

This reveals a fundamental issue in developing AI products for education: “Who defines the needs?” The needs imagined by tech companies based on their technical capabilities often diverge from what actually happens in classrooms. Ryan, being a teacher himself, knows which areas most need support and which boundaries must not be crossed. A user-driven, rather than technology-developer-driven, product design approach is rare in educational AI—but it may be the path to truly useful educational tools.

In February 2025, Ryan approached the Learnovate Centre with his idea to initiate formal collaboration. At the same time, Diotima secured a €500,000 commercialization grant from Enterprise Ireland. Learnovate assigned two AI postdoctoral researchers to handle technical development for the project. The initial platform was launched in September 2025, followed by two large-scale testing rounds, and the company was spun off as an independent entity in 2026. The entire journey from partnership launch to independent company formation took one and a half years.

Learnovate’s funding is primarily provided by Enterprise Ireland and IDA Ireland, the core government agencies responsible for driving industrial innovation and attracting foreign investment in Ireland. In 2024, Learnovate secured €9.6 million in new funding to advance research in future learning and skills development. The center offers comprehensive support to resident projects, spanning research and development, testing, and commercialization.

Learnovate has a global network of industry partnerships, including international giants such as Zoom, Cisco, and Mastercard, as well as numerous educational institutions, schools, and vocational training organizations. Projects incubated by Learnovate gain direct access to real-world business needs and educational contexts from the outset, enabling them to receive feedback and testing opportunities from potential customers during the product development phase. Diotima’s testing program is conducted within this network, with secondary and vocational education institutions directly participating in product iteration.

Diotima’s inaugural CEO is Jonathan Dempsey, an expert in commercializing edtech, who previously served as CEO of Digitary, an edtech company supported by Enterprise Ireland. Ryan himself serves as Chief Product Officer and Learning Lead. The technical team includes development engineer Daniel Fernandez and AI engineer Dr. Long Mai. The project has engaged Dr. Eoin Lane as a governance advisor, formerly the Global Head of AI and Data Science at BNY Mellon, specializing in AI regulatory compliance. This configuration demonstrates that Diotima is a comprehensive product with a clear commercial focus, compliance prioritized from the outset, and equal emphasis on educational expertise and technical capability.

In February 2026, Learnovate launched the industry initiative titled "Responsible AI for Learning," abbreviated as RAIL. Dempsey, serving as the Business Lead of Diotima, chaired the initiative, which included participation from schools, higher education institutions, vocational education organizations, teachers' unions, and representatives from Ireland’s Department of Education. The initiative aims to establish industry standards in educational AI, addressing regulatory requirements, risk identification, and implementation pathways.

The core of this mechanism is that universities serve as commercialization incubators, deeply involved and providing full-cycle support. Academic research and commercial transformation are a continuous process, with universities as active participants.

Diotima integrated regulatory requirements such as the EU AI Act, GDPR, the Digital Services Act, and the European Accessibility Act into its product architecture from the design phase.

The EU AI Act imposes strict limitations on the use of AI in educational contexts. Education is classified as a high-risk sector, and any AI system used to assess learners, influence educational decisions, or provide learning recommendations must comply with requirements for transparency, explainability, data protection, and human oversight. General-purpose large models like ChatGPT face compliance challenges in educational settings due to unclear data flows, unexplainable algorithmic black boxes, difficulties in ensuring student privacy, and ambiguous mechanisms for teacher supervision of AI outputs. As a result, many schools and educational institutions remain cautious about adopting general-purpose AI tools.

Diotima makes compliance part of its product DNA. The platform is designed with data processing workflows that meet the strictest regulatory standards: student data is processed locally and never uploaded to external servers; algorithmic decision-making is fully traceable, allowing teachers to see the basis for AI-generated feedback; and the system never makes final decisions—all evaluations must be reviewed and approved by teachers. These design choices enable schools and educational institutions to use the platform with confidence. In the European market, regulatory requirements will only grow stricter, making this advantage increasingly significant.

AI regulatory compliance experts like Eoin Lane are deeply involved in product design. He has handled numerous AI compliance issues in the financial sector at BNY Mellon. The financial and education sectors share similarities in terms of regulatory rigor and risk control requirements.

The importance of this path choice lies in the fact that, in sensitive areas like education, any restriction or requirement for remediation due to compliance issues could instantly erase all previously accumulated users and data. By treating compliance as a design constraint from the very beginning, you gain long-term sustainability and trust in the B2B market. Schools and educational institutions are extremely cautious when selecting tools, prioritizing not only functionality but also compliance, security, and controllability.

This strategy also creates industry barriers. The requirements of the EU AI Act are highly specific and technical, requiring significant resources and time to fully comply. Large companies may have the capability but may not be willing to invest in custom development for a relatively niche market like education; smaller companies may want to enter but lack sufficient resources and expertise. Diotima has crossed this barrier in advance by partnering with Learnovate and Trinity and hiring professional compliance advisors.

As regulatory frameworks continue to mature—particularly in strictly regulated markets such as Europe and North America—compliance capability may become a more critical competitive factor than technical ability. Technology is the entry barrier; what truly determines whether a platform can establish a foothold in the market is its ability to pass regulatory scrutiny and earn the trust of schools and educational institutions.

Diotima is currently in its early stages. Public information has not disclosed user numbers, revenue figures, or specific market expansion plans. It has transitioned from a university project to an independent company, secured its first round of funding, assembled a team, and established a basic framework. Whether it can truly establish a foothold in the market remains to be seen.

Issues to observe include: whether this requirement is sufficiently critical, whether teachers are willing to change their existing work habits to adapt to the new tool, whether schools are willing to pay for such a platform, and whether the quality of AI-generated assessments and feedback truly meets teachers’ standards. The education market is characterized by long decision-making cycles, complex procurement processes, and high switching costs; a B2B edtech product often takes several years to move from pilot testing to large-scale adoption.

But this case provides several clear reference points:

The right to design educational tools: should it be driven by tech companies based on technical capabilities, or by teachers based on instructional needs? Diotima proves the latter is feasible.

The role of universities in educational innovation: purely research institutions, or incubators for industry-academia collaboration? Trinity and Learnovate embody the latter, with deep, end-to-end involvement.

Is compliance a burden or a competitive advantage? In a phase where AI technology is rapidly advancing but regulatory frameworks are still evolving, embedding compliance into product design from the outset—though it may reduce some flexibility—could be the right strategy for educating the B2B market.

For China’s education sector, the regulatory environment, educational system, and market structure differ from those in Ireland, but some underlying principles are shared: AI tools in education must genuinely serve teaching needs, product development requires deep involvement from education professionals, and compliance and security must be considered from the outset. While China’s regulatory environment differs from that of Europe and the U.S., requirements around data security, algorithmic transparency, and educational ethics are also continuously increasing.

Whether Diotima will succeed, establish a foothold in the European education market, and truly integrate the concept of “responsible AI” into its products—these answers will be determined by the market. Yet, it also represents a possible path for the development of educational AI—one distinct from the “all-in” strategies of big tech firms and the aggressive expansion of startups.

Original link:

Diotima, an AI education technology company founded by educators, has been spun out from Trinity.