2025-09-22 AI as Coach

Assessment Prep with AI

As business advisors we are frequently asked how AI can be incorporated into business processes.  Back in 2023 we saw Oracle incorporate Generative AI into their applications in 2024.  The use cases were really good, but filled in work that was already within the ERP processes.  We thought there was a mode of interaction with AI that was more akin to having a coach or advisor.  We have been able to test this view in the setting of an application to be a detective with the Metropolitan Police.  It has been a way to have an AI elicit the best from us rather than have it do something on our behalf.  In an age where we have genuine fear that AI will replace workers, this sheds some light to how AI might contribute to human flourishing.

Stage One: Situational Judgement Test

We used it to explain assessment criteria, to analyze our responses and suggest improvements.  It was interesting to see that while we could get it to create model responses, and that it obviously scored its model responses very highly, we did not like its model answers.  There is still something human that was missing when you are looking for a genuine human judgment on a truly human situation.

Stage Two: Online Assessments

The second stage involved a set of more in-depth online assessments, including:

• An in-tray exercise

• A written report

Here, AI played a more active role in rehearsal and preparation. For example:

Simulation of Conversations

We could play with different tones and character traits, increasing the amount of stress that our AI Witness was under and forcing our prospective police officer to respond.

AI Feedback

AI tools were used to give real-time critique, such as flagging when responses sounded overly formal, reminding the candidate to acknowledge emotions before moving to problem-solving, or suggesting smoother phrasing for written exercises.

Building a Custom AI Coach

With no technical setup needed, the candidate could paste in draft answers or case study prompts and ask the AI to challenge them with variations. The focus wasn’t on having AI produce final answers, but on developing the candidate’s own ability to think, structure, and respond under time pressure.

Stage Three: Final In-Person Assessments

The last stage, currently in preparation, includes:

• A face-to-face interview

• A roleplay exercise

• Another in-tray exercise

Here our emphasis shifts away from AI preparation and onto live performance. AI can still be used in advance to rehearse roleplays or practise interview questions, but assessors will be looking for calm, authentic, and professional responses in real time. The key is in the simulation of similar live situations with AI as a bouncing board, which will enable a candidate to be more fluid,practiced, and natural when encountering similar situations on the actual assessment day. 

Additional Technique: Using AI Image Analysis to Review Handwriting

As part of our preparation for the Metropolitan Police assessment, we also experimented with AI image analysis to review our handwritten responses. This exercise helped us understand how handwriting itself reflects focus and composure under pressure.

We began by writing short responses to the report-style prompts under timed conditions, then photographed the pages and uploaded them to an AI handwriting-analysis tool. These tools, similar to those used for document verification, can detect variation in spacing, line consistency, and stroke pressure. In our case, the AI highlighted points where our handwriting became tighter and slanted  typically when we were rushing an answer.This simple process revealed more than we expected. It gave us a picture of where stress appeared on the page, allowing us to practise writing at a steadier rhythm. The aim wasn’t to produce perfect handwriting, and to maintain clarity and calm presentation throughout.