Hi [PO Name],
I wanted to share my observations and feedback after working with the standardized Copilot prompt for KB document refinement.
Observations:
The prompt is structured around five key metrics and performs a thorough scan across each section. It provides detailed inputs along with a clear “safe/unsafe” verdict and reasoning, which helps identify gaps and build confidence while making improvements.
It also enforces consistency across documents and enhances them by expanding procedures into more detailed, step-by-step guidance.
Feedback:
One challenge is the increase in content length. For end users, this may result in more information than needed when they are looking for quick, specific answers. However, if the long-term plan is to integrate with Agentive AI, the structured and detailed format should still be beneficial.
Additionally, each metric requires evaluation and often rewrites. While this improves quality and confidence scores, it does add effort and may require multiple iterations.
The requirement to include ownership/escalation details may not be applicable for all KBs and could be made more flexible.
Time Taken:
On average, it takes about 30 minutes to process a document, including iterations to incorporate suggested improvements and achieve a “safe” verdict, which is not always met in the initial pass.
Overall, the prompt is effective in improving quality, consistency, and completeness, though some optimization around content length and flexibility would enhance usability.
Re
gards,
[Your Name]
Hi [PO Name],
Observations:
The prompt evaluates documents across 5 metrics and provides detailed inputs with a clear safe/unsafe verdict and reasoning. This helps identify gaps quickly and improves confidence. It also ensures uniformity and enhances documents with clearer step-by-step procedures.
Feedback:
Content length increases significantly, which may impact quick readability for users. However, it works well if aligned with future Agentive AI use. Iterations are often needed to improve scores, and ownership/escalation details may not be relevant for all KBs.
Time:
~30 minutes per document, including iterations to address gaps and achieve a “safe” verdict.
Overall, it improves consistency and quality, with some trade-offs in length and iteration effort.
Regards,
[Your Name]