TL;DR AI ambition is growing quickly in higher education, but skills gaps, governance questions, legacy systems and unclear ROI are making it harder to move from pilots to everyday use. Technology leaders need to focus on the foundations that make AI safe, useful and easier to scale across the institution.
NashTech’s survey data shows that 88% of education technology leaders expect to adopt or increase their use of AI within the next two years. Generative AI pilots are already being explored across teaching, learning, assessment, student support and operations. The challenge now is turning those early ideas into trusted, joined-up services that make a real difference for staff, students and the wider institution.
Sector guidance reflects the same shift, with Jisc’s AI maturity toolkit encouraging institutions to move from early experimentation to more embedded, mature AI adoption.
For technology leaders, this means a change programme that needs the right people, processes and safeguards around it.
In this article, we look at some of the challenges HE technology leaders should focus on now, to help move AI from pilot to campus-wide impact.
Many existing technology strategies were not designed with AI in mind.
For HE leaders, that creates a few practical issues:
HE technology leaders need to move away from asking where they can ‘try AI’ and instead focus on how AI can support institution-wide change. That means understanding how their technology roadmap, data strategy, governance and institutional goals can enable AI to scale safely and effectively.
Many universities are already refreshing their technology strategies. NashTech’s survey data shows that:
This is a positive step, but strategy only helps if it leads to clear action. AI is hard to scale when pilots sit on their own or technology decisions are made in isolation. Leaders need to understand where AI can create value, which systems it needs to connect with, and which platforms can support future change.
Action for technology leaders: review your technology roadmap against your priority AI use cases. Identify where your systems, platforms or partner support need to change before pilots can move into everyday use.
A lack of internal AI architecture and engineering expertise is one of the clearest barriers to scale. NashTech’s survey data shows that 42% of education technology leaders cite this as a challenge for generative AI, while 37% cite it for agentic AI.
This creates a practical delivery gap. Teams may see the opportunity, but may not yet have the specialist skills or capacity to design, build, govern and support AI safely once it moves beyond a pilot.
Action for technology leaders: map the skills you have today against what is needed to design, integrate, test, secure and support AI services. Use the gaps to guide hiring, training and partner decisions.
AI will not deliver much value if it sits outside the systems staff and students already use. Difficulty integrating AI with legacy systems is already a core blocker: 42% cite this challenge for agentic AI and 36% for generative AI.
For universities, integration is more than a technical issue. If data is spread across learning platforms, student systems, service desks and reporting tools, AI use cases become harder to scale, govern and measure.
This is also becoming a sector-wide priority. UCISA is working on the National Student Data Model, a framework designed to map student data structures and accessibility in a more consistent way across UK higher education. Based on the Dutch OEAPI model, it aims to support greater uniformity and standardisation so student data can be structured, understood and moved more easily between institutions. For technology leaders, this matters because integration readiness is no longer only about internal systems. It is also about preparing for a more standardised data environment across the sector, including the Lifelong Learning initiative due to commence in January 2027, which adds a funding and policy driver to the need for better student data portability.
Action for technology leaders: review the systems, data flows, APIs and student data structures that your priority AI use cases rely on. Consider how emerging sector frameworks, such as the National Student Data Model, could shape future integration and data portability requirements. Then prioritise the fixes that will provide the most value and reduce the biggest delivery risks.
Governance, compliance and safeguarding are central to AI in higher education. Institutions need to think carefully about student data, academic integrity, responsible use, privacy, transparency and safety before AI becomes part of everyday workflows.
NashTech’s survey data shows that data governance and compliance uncertainty is a recurring barrier, cited by 41% of education technology leaders for agentic AI and 39% for generative AI. The risk is that governance becomes a blocker if it is only considered at the end.
Action for technology leaders: set up a simple AI governance process before pilots begin. It should cover acceptable use, data access, human oversight, security review and sign-off criteria.
AI services need more than a promising idea. They need reliable infrastructure, clear ownership, monitoring and support. Without these foundations, pilots may work well in a small test but struggle when they are used more widely.
This is already visible: 41% of HE technology leaders cite limited access to scalable infrastructure or MLOps capabilities as a barrier for generative AI, while 33% cite it for agentic AI.
Action for technology leaders: agree what “ready for wider use” means for AI services. Include monitoring, escalation, access control, resilience, cost management and regular performance checks.
Budget pressure and unclear ROI can slow progress, especially where teams are already stretched. NashTech’s survey data shows that budget constraints and unclear ROI are major inhibitors, cited by 44% for agentic AI and 41% for generative AI.
Technology leaders need to show practical value, not just technical possibility. That could mean reducing admin for staff, improving student support response times, helping with academic feedback, automating reporting or improving intervention workflows.
Action for technology leaders: define the value case for each AI initiative before delivery starts. Set a baseline, success measure, owner and route to wider use.
Moving from pilot to campus-wide impact means shifting from standalone tools to joined-up workflows. The value is not in the AI tool alone. It comes when AI is built into the processes, data and decisions that matter. For example, an AI assistant that helps triage student enquiries only becomes useful if it works from trusted data, follows approved policy and supports staff safely.
This needs joined-up leadership across IT, academic teams, operations, legal, data governance, safeguarding and student services. EDUCAUSE’s 2025 AI Landscape Study highlights the importance of strategy and leadership, policies and guidelines, use cases, workforce development and the institutional digital AI divide. Jisc’s framework also shows that AI maturity depends on skills, technology, governance and data readiness working together.
In practice, universities need a clear way of managing AI, not just a technology stack. That should include sponsorship, value cases, responsible AI governance, workforce development, adoption planning, data and integration readiness, and ongoing measurement.
What HE technology leaders should do next
To close the gap between AI ambition and real-world impact, focus on five practical priorities: