The case for the omnichannel AI assistant

Most institutions still rely heavily on traditional learning management systems. They’re functional but they’re also often static, desktop-centric and disconnected from the digital habits of Generation Z students.

At our university, we in the department of applied social sciences identified that to truly bridge this gap and modernise the student experience we needed a platform that lived where the students live. To achieve this, we developed the Virtual Assistant TIMS.

• GenAI as a research assistant
• An insider’s guide to how students use GenAI tools
• WhatsApp and campus trails: supporting students in building peer-support networks

The virtual assistant is not merely another application; it is a smart omnichannel chatbot integrated directly into WhatsApp. It represents a small but critical step towards the ideal future platform: one that is seamless, ubiquitous and globally accessible. 

Here, we’ll outline our implementation strategy, the challenges regarding academic integrity we navigated, and practical takeaways for educators across disciplines.

Dissolving boundaries: the omnichannel approach

What would an ideal future platform look like? We believe frictionless access is the most important thing. Our initial analysis found that requiring students to log into a separate portal for every minor query created a barrier to engagement.

Embedding the virtual assistant into instant messaging platforms that students already use made it ubiquitous. It dissolves the boundary between being in class and daily life. For educators looking to replicate this, our advice is to prioritise integration over destination. Rather than building a stand-alone tool, look for ways to embed AI assistance into the students’ existing digital ecosystem.

The assistant exemplifies this by offering real-time lecture summaries and accepting audio-based queries directly through the messaging platform. This aligns with students’ natural communication habits, allowing them to digest complex lecture material or ask questions on the go using voice notes. This seamless transition between synchronous and asynchronous modes extends the learning window significantly beyond office hours.

From oracle to mirror: addressing cognitive laziness

A primary concern among faculty is that AI assistants could foster cognitive laziness or facilitate plagiarism. This fear is valid: if the bot answers the question, will the student stop thinking?

Our experience with the virtual assistant suggests that this outcome is determined entirely by the design of the interaction. We designed the tool not to provide answers but to provide “feedforward”. The assistant acts as a mirror rather than a writer.

This is achieved through rubric-based assessment feedback and video presentation assessments. Instead of simply generating content, the assistant analyses student submissions – whether text, audio or video – against specific grading criteria. It identifies gaps and offers critiques based on the rubric, ensuring the student still does the cognitive work of revision.

For colleagues in other departments, we recommend programming the AI to function as a critical friend. The goal is to configure the tool to critique student input rather than complete the task for them. By structuring the interaction this way, the assistant stimulates the student to engage in iterative revision. It turns the process into a dialogue, improving the final quality of work while preserving the student’s agency.

Navigating hallucinations

Implementing the assistant was not without operational hurdles. The most significant technical challenge was the hallucination problem – where the AI confidently presents incorrect information or generates content out of context.

To solve this, we moved away from an open-ended model to a closed AI environment. We mandated that the AI’s knowledge base be strictly limited to teacher-provided lecture handouts and verified subject materials. By grounding the AI solely in the course curriculum, we ensure that every response is accurate, relevant and contextually aligned with what is being taught in the classroom.

Furthermore, we integrated the tool into the live learning experience. We encourage teachers to use the assistant during class for real-time activities and Q&A sessions. This creates a dynamic feedback loop: as students and teachers interact with the tool together, we can constantly monitor the output. By listening to feedback and facilitating open discussions about the AI responses, we ensure the platform remains a helpful, accurate and trusted companion for learning.

The Virtual Assistant TIMS is a small step towards Education 4.0, our university’s AI-powered, student-centred teaching and learning model, but it signals a significant shift in how we conceptualise the campus. The ideal platform of the future is not a destination students must visit; it is an intelligent companion that travels alongside them.

Rodney Chu is senior lecturer and Charles Woo is project fellow, both in the department for applied social sciences at the Hong Kong Polytechnic University.

If you would like advice and insight from academics and university staff delivered direct to your inbox each week, sign up for the Campus newsletter.