Skip to main content

AI IT Support Agent (Københavns Universitet)

An AI support chatbot for KU staff and students, grounded in the university's own ITSM knowledge base.

Copilot Studio Agent Copilot Studio Power Platform Dataverse AI Builder Power Platform Pipelines Ivanti ITSM High 8 weeks
About this agent

Copilot Studio agent that helps internal users at Københavns Universitet get precise, fast IT support. It is grounded on KU's own Ivanti ITSM knowledge articles (migrated into Dataverse and enriched with an AI-generated category tag across 15 fixed categories). A confidence-score check validates every answer. Delivered as a PoC with incremental releases, side-by-side pairing with KU's internal developers, and a roadmap from PoC through full production rollout across the entire university.

Key capabilities

What this agent does

  • Answers KU IT support questions grounded in migrated Ivanti ITSM knowledge articles (Dataverse-hosted, AI-categorised into 15 fixed categories).

  • Confidence-scored generative answers composed via AI Builder prompts in the tone of a KU support engineer.

  • Three-environment ALM (Dev unmanaged / Test managed / Prod managed) with Power Platform Pipelines solution deployment — Dev → Test → Prod.

  • Planned escalation flow to human KU-IT support (phase 2) when confidence is low or the topic sits outside the curated knowledge base.

  • Side-by-side delivery with KU's own developers (Kenneth, Simon, Jonas) — building KU's internal capability to extend and maintain the agent.

Business value

What you can expect

Measured as deflection rate of routine IT questions from the human queue and cost-per-resolved-question.

Measured as improved answer consistency grounded in validated internal knowledge — with the open policy question of whether webgrounding / LLM general knowledge is permitted explicitly flagged.

Measured as reduced routine-question load on the central KU-IT team, freeing them for higher-complexity work.

Measured as time-to-answer for staff and students (seconds vs. waiting on a ticket) across the full university, not just one faculty.

Target audience

Who benefits

KU Staff / Student KU-IT Support Engineer KU Automation & New-Tech Lead (Kenneth) Platform Developer / Architect (Simon, Jonas)