As universities increasingly embrace AI tools to enhance research and education, I’ve been exploring how to create specialised Copilot agents that can genuinely support PhD students and supervisors. Here’s how I approached building two distinct agents tailored for academic environments, particularly within the context of a modern Scottish university.

The Challenge: Meeting Specific Academic Needs

PhD programmes present unique challenges that generic AI assistants often struggle to address effectively. Students need support with complex research processes, academic writing conventions, and navigating institutional requirements. Supervisors, meanwhile, require tools that help them manage multiple students whilst maintaining quality standards and meeting university obligations.

Rather than relying on one-size-fits-all solutions, I decided to create purpose-built agents with specific expertise and behavioural guidelines.

Agent 1: The PhD Student Assistant

The first agent focuses entirely on supporting doctoral students throughout their research journey. Key capabilities include:

Research and Writing Support

  • Literature review assistance and synthesis
  • Academic writing feedback and style guidance
  • Citation management and formatting help
  • Thesis chapter development support

Technical and Methodological Guidance

  • Data analysis assistance and interpretation
  • Research methodology recommendations
  • Statistical support and visualisation help
  • Research tool troubleshooting

Academic Navigation

  • Understanding of PhD milestones (transfer vivas, annual reviews)
  • Time management and project planning
  • Academic integrity awareness
  • Conference and publication guidance

Knowledge (linked resources)

The agent is programmed to encourage critical thinking rather than providing direct answers, maintaining academic rigour whilst being genuinely helpful.

Agent 2: The Supervisor Assistant

The second agent addresses the unique challenges facing PhD supervisors, particularly around managing multiple students and administrative responsibilities:

Student Progress Management

  • Milestone tracking and deadline monitoring
  • Supervision meeting planning and agenda setting
  • Progress assessment and intervention identification
  • Documentation and record keeping

Administrative Coordination

  • Graduate school requirement management
  • Ethics approval processes
  • Funding and studentship administration
  • External examiner coordination

Quality Assurance

  • Consistent feedback standards
  • University policy compliance
  • Research excellence alignment
  • REF preparation support

Knowledge (linked resources)

Tailoring for Context: The Scottish University Perspective

Working within Glasgow Caledonian University’s context, I ensured both agents understand:

  • Scottish higher education structures and funding landscapes
  • University-specific procedures and policies
  • Local research partnerships and opportunities
  • Sector-specific career pathways

This contextual awareness makes the agents far more practical and relevant than generic alternatives.

Key Design Principles

Specificity Over Generality Rather than creating one broad agent, I developed two focused tools that excel in their specific domains. This approach ensures more relevant, actionable support.

Cultural and Linguistic Appropriateness Using British English and understanding UK academic conventions makes the agents feel natural and appropriate for their intended users.

Institutional Awareness Building in knowledge of specific university procedures, deadlines, and resources makes the agents genuinely useful rather than theoretical.

Ethical Boundaries Both agents are designed to support learning and development rather than replace critical thinking or academic judgement.

Implementation Tips

When building similar agents, consider:

  1. Start with user interviews – understand the specific pain points your agents need to address
  2. Be institution-specific – generic advice is often less valuable than contextual guidance
  3. Define clear boundaries – ensure agents support rather than replace human expertise
  4. Test iteratively – refine based on actual usage patterns and feedback
  5. Maintain academic integrity – build in safeguards that encourage proper academic practice

Looking Forward

As these agents evolve, I’m exploring how they might integrate with existing university systems and support broader research culture development. The goal isn’t to replace human interaction but to free up time and mental space for the high-value conversations and thinking that make PhD programmes transformational.

Building effective AI agents requires understanding not just the technology, but the human context in which they’ll operate. For universities considering similar projects, start small, think specific, and always keep the end user’s actual needs at the centre of your design process.

If you would like a copy of the actual instructions I used please get in touch via email or LinkedIn.