I'm currently building a plumbing estimator system that triggers off a web-hook from a form that gets filled out with basic information and the blueprints for claude to understand at least the basic scope of the blueprint or the project. then it continues to go through a series of code nodes, https nodes, a human review area for any flagged issues or questions the ai detected, then an ggle sheet for data logging, and more other nodes until it submits a PDF report to the users email. my issue currently is with accuracy of pricing and glitches/inaccuracies that is reported in the PDF report (the estimate) in the end. if anyone is the construction ai industry, could yall help? if anyone has any idea a well, im open to any sort of help or advice.
Thanks! maybe you can help somehow haha