Read the business request
Sign in to save your workRead the request as a governance practitioner. Your task is to test whether the case for PolicyAI is supported by evidence, not to validate the solution.
“PolicyAI is popular and saves HR time, but we have had two incidents. Should we keep it running, and what must change?”
PolicyAI is a chat assistant on the staff intranet. It searches Meridian's HR document library, passes the most relevant passages to a third-party large language model and returns an answer with a cited source.
Staff type a question about pay, leave or policies. The system looks up documents and asks an AI model to write an answer. If it picks up the wrong document, or one it should never have seen, people get wrong advice or see private information.
- Six evidence documents
- Five fictional staff queries
- Simulated answers with cited sources
- A sample of answers to test against current policy
- Answer routine HR questions instantly
- Cut HR helpdesk tickets by 40%
- Give consistent answers across 38 sites
- Free HR advisers for complex cases
Your first challenge
Write your opening questions before you look at the evidence.
Think about purpose, data, affected people, evidence, accountability and the human role…
Answers save automatically and sync to your account when you are signed in.
Every organisation, person and figure is fictional. Do not enter real personal, customer or patient information.