A Work in Progress
Our AI code assistant is up and running, and I hope that those of you that are early adopters are receiving value from it! As with all things technical, nothing is ever 100% at the get-go, it takes real-world feedback to identify any shortcomings and gradually improve things.
As part of my own process, I gave it a test on 'allowable exposed openings' based on limiting distance. Anyone that has done this type of evaluation knows that it's a bit of a chore. My first test was based on two wall faces of differing areas at different limiting distances, and it scored a 50% - Wall 'A' is correct but Wall 'B' came in more conservatively than it should have.
Our next step is to evaluate the process that the AI used to determine its values, and provide some additional training to correct the error. I feel like if we can resolve this issue, most others will be easier to address :).