How to build AI agents to help us for monitoring the platforms

[sidram.m]

Build an AI-powered RPA Operations Assistant that continuously monitors all bot runs across different time zones. Detect failures in real time, identify the root cause, and automatically resolve common issues wherever possible using AI agents. For issues that cannot be resolved automatically, generate a clear diagnosis, recommended actions, and immediately notify the support team with the relevant details. suggest me how we can build this one? 

[david.l.morris]

I want to provide you some constructive feedback here. This question is not well crafted. When asking a question on any given forum, the original poster typically needs to explain the full context, what they've tried, what did not work, what specific challenges they've faced, examples, etc.

What it appears to me that you're doing is trying to get other people to do your job. I'm not saying that's what you're actually doing, just that it gives that appearance.

The topic you've brought up is extremely complicated. No one is going to respond with an answer that is going to help you because most people do not want to have to guess all of the possible scenarios that you may be referring to as well as guess the things you may have tried or looked into.

[Denis__Dennehy]

I have seen similar built over many years by customers - controllers/ops teams should not have to look at Blue Prism Control Room for anything other than making changes to run times - instead they should have dashboards to monitor a wide variety of processes (created using data gateways) and they should have alerts that inform them of system unavailable failures (a major failure where a process cannot - not occassional exceptions) - there are a number of alert solutions in best practice training and in the digital exchange.

The requirement for some AI to understand issues and automatically resolve them is a step further - I am guessing that could be possible but every customer's solution is different - with different logic, different systems, different solutions - therefore an AI controller robot would probably need to be something you would have to build and train yourself.  For example, error x usually means y and is usually fixed by restarting the virtual desktop.  An agent could potentially be trained to do that but it would not be something I would expect from Blue Prism.