topic Re: Blue Prism Data Gateway → Session Logs → External Python/Local LLM for AI-Based RCA in Product Forum

Blue Prism Data Gateway → Session Logs → External Python/Local LLM for AI-Based RCA

sidram.rpa — Mon, 21 Sep 2026 15:14:57 GMT

Whether Data Gateway can push the required session-log granularity directly to your Python endpoint, rather than assuming it can. That distinction will determine whether your architecture should be:
Data Gateway → API → LLM
or
Data Gateway → Database → Python → LLM.

please do suggest me on the same?

Re: Blue Prism Data Gateway → Session Logs → External Python/Local LLM for AI-Based RCA

steven.boggs — Thu, 24 Sep 2026 13:19:19 GMT

I'd suggest you expand on your question here so others could provide some suggestions. For example, what is the end-goal of this type of implementation? In the title of this thread you mention "for AI-based RCA" -- what type of Root Cause Analysis would this project provide, for what types of issues, and concerning which kinds of hardware or software faults? Would this project just provide basic analysis of a fault or exception in BPE session logs, or suggest solutions for preventing them as well? Can you describe the "Python endpoint" you mention further?

As much detail as you can provide would be helpful to determine if we might have something ready-made we could suggest in our Digital Exchange, or if another Blue Prism user has implemented something similar and could provide their feedback.

Re: Blue Prism Data Gateway → Session Logs → External Python/Local LLM for AI-Based RCA

sidram.rpa — Fri, 25 Sep 2026 12:10:51 GMT

I am developing an AI-powered RPA monitoring solution to reduce the manual effort involved in monitoring 50+ Production bots.
AI agents will analyze bot logs, identify failures, perform RCA, and generate actionable alerts.This will reduce repetitive monitoring and troubleshooting efforts while helping the team respond to issues faster.

Blue Prism logs can be collected through Blue Prism APIs or Data Gateway, and I would like to explore how to utilize these capabilities effectively.

The goal is to determine the most suitable approach for feeding operational data into AI agents for monitoring, RCA, and proactive alerting.