Words by Zach Pendleton, chief architect, Instructure
The first wave of generative AI is settling into higher education. But while some universities are still managing the implications of chatbots in the classroom, balancing their support for the learning process and weighing the risks to assessment integrity and academic standards, other institutions are already incorporating agentic AI.
They are transitioning from exploring simple queries and prompts to a more sophisticated form of automation that is quickly being integrated into the ecosystems of leading universities worldwide.
Agentic AI refers to systems capable not only of generating content but also of reasoning, planning and executing multistep tasks. This new technology is poised to reshape the way universities build and manage their digital infrastructure to scale personalised learning.
Universities are past the pilot stage with AI
Early explorations of AI in education have demonstrated its potential to teachers and students, revealing the implications of using tools that do not always work well together. Inconsistent experiences for students, complexity in data management and additional burdens on administrators tasked with linking systems that were never designed to integrate, are just a few of the challenges universities are facing.
The speed at which AI is evolving demands strategies that go beyond layering AI features onto outdated systems. It calls for institutions to integrate both generative and agentic technologies directly into their learning environments. That means designing courses and assessments that take account of how these tools are used, viewing them not as threats to be policed but as opportunities for structured, outcome-driven engagement.
It also means setting a far higher bar for transparency. Universities must use virtual learning environments (VLE) that ensure student data is not being diverted to train external models and that information flows within platforms can be explained clearly to staff and students.
Agentic AI won’t work where there’s fragmentation
Most legacy learning platforms were created to support straightforward, deterministic processes rather than dynamic workflows requiring context from multiple databases, planning dynamic sequences of actions or adapting to individual users.
Agentic AI cannot function effectively if the infrastructure beneath it is closed, fragmented or outdated. What universities increasingly require is agent-ready architecture, meaning unified data access under institutional control, robust and transparent application programming interfaces that support emerging standards, and flexible interfaces designed for both human and machine use.
Institutions that rely on self-hosted virtual learning environments often started down that path with the hope that they would have more control over their future, but they’re now at risk of a future full of patching legacy architectures with superficial AI features and doing a disservice to students. In contrast, those that modernise their VLE to support orchestration across systems will be better placed to offer safe, consistent and innovative teaching and learning experiences.
The arrival of agentic AI represents the potential to enter a new era of digital learning. Using the right VLE will be critical for a successful implementation. Poorly coordinated digital strategies risk further fragmentation, rising costs and greater governance difficulties.
What education will look like with agentic AI
For students, the ability to translate, summarise or navigate personalised learning pathways within the learning management system itself – without relying on external apps – reduces friction and offers greater protection for privacy.
For faculty, agentic AI may help align rubrics, highlight at-risk learners by detecting patterns of engagement, and propose tailored interventions.
For administrators, routine queries and processes can be handled automatically, freeing staff time for more strategic work that helps institutions make data-informed decisions.
For all, agentic AI is like having a personalised assistant on call to handle your most frustrating tasks. But it only works when safely and properly integrated with your existing tools.
A future with agent-friendly VLE
The real opportunity lies not in adopting isolated AI-powered tools but in creating the conditions for AI agents to operate across the institutional ecosystem without compromising trust or control.
According to the Times Higher Education 2024 Digital Maturity Index, 80 per cent of higher education institutions globally report having access to a VLE (also known as an LMS). It is clear that the VLE has become a foundational requirement for modern teaching and learning. The capabilities of the VLE, and utilising it to its full potential, are what really empower institutions’ digital maturity advancements.
In this context, only universities that make VLE modernisation part of their pedagogical strategy will be able to build an ecosystem with the necessary infrastructure to offer agent-friendly, cloud-based systems, ultimately helping to place themselves at the forefront of the next stage in higher education’s digital transformation.
Find out more about Instructure.
comment