Bringing Hybrid, Agent-based AI to Higher Education

Higher education institutions are actively seeking new ways to improve student services while considering practical limits around staffing, budget, and infrastructure.

Traditionally, student services have been provided through a mix of websites, printed materials, and in-person consultations. However, as students now expect quicker, more engaging, and self-service options, colleges and universities are reassessing how they can enhance service accessibility on a larger scale.

With a growing interest in AI-driven solutions, institutions are exploring conversational platforms as a means to deliver digital services. These systems are designed to efficiently process requests, provide information, and guide users through procedures using everyday language. Implementing this model at scale presents broader IT challenges, particularly regarding how AI-driven services are delivered and the best environments for different workloads.

The evolution of student services in higher education.

A notable trend in AI delivery is the use of hybrid architectures that combine local processing on advanced AI-enabled devices with cloud-based AI services reliant on large language models. For instance, the ibl.ai platform pairs agent-driven design with the ability to execute certain AI tasks locally, particularly on devices like Microsoft Surface devices with NPUs, while utilising cloud services for broader institutional needs. This setup allows institutions to dynamically decide where processing should take place, influencing performance, data management, and operational costs.

As outlined in Microsoft’s analysis of small language models for retrieval-augmented generation, SLMs can be effective for particular tasks when their operation and deployment context are well-defined. Here are a few situations where they shine:

• Offline or local environments: Ideal for cases where local processing is vital due to connectivity issues, policy restrictions, or architectural needs.
• Latency-sensitive interactions: Important where the placement and execution of the model influence the design.
• Budget-constrained scenarios: Useful for organisations that want to limit cloud processing costs by selectively handling certain tasks locally.
• Resource-limited environments: Key when the model size and deployment footprint must be considered.
• Task-specific workflows: Best when smaller models are tailored to specific use cases rather than relying on general-purpose models.

It’s crucial to recognise that decisions around where and how AI operates depend on institutional design, governance needs, and technical limitations—not solely on the AI models themselves.

The ibl.ai platform showcases how hybrid AI can function effectively within higher education. Working alongside the Surface development team, ibl.ai has facilitated on-device processing and created the student experience pack. This is a collection of special AI agents designed for use on Surface Copilot+ PCs equipped with NPUs. These agents assist with common student activities like accessing the Study Hub, Campus Connect, and Career Launchpad, as well as services related to Surface devices, such as onboarding and support.

This collaboration illustrates how services can be adapted to operate either online or offline, based on institutional needs and conditions. Typically, these agents connect to cloud services for comprehensive access, but in offline situations or under certain usage rules, they can function completely independently on the device.

As AI becomes a bigger part of student interactions, colleges and universities must navigate competing demands. Students expect fast, conversational support, while institutions strive to meet these expectations responsibly, without compromising governance or sustainability. IT teams require systems that can be flexible across different devices, networks, and deployment models.

For students wanting easy conversation-driven access to services, institutions need an effective delivery method—one that scales, maintains control, and employs hybrid AI to enhance value without negative consequences.

FAQs

1. What are student-facing AI services?

Student-facing AI services refer to digital tools that assist students with various academic and administrative tasks through conversational interfaces, enhancing accessibility and immediacy.

2. How do hybrid AI systems work in education?

Hybrid AI systems combine local device processing with cloud-based services to manage tasks effectively while considering factors like cost, speed, and data privacy.

3. Why is AI important in higher education?

AI is crucial in higher education as it can streamline processes, offer personalised support, and meet student needs promptly, fostering a better learning environment.