Skip to main content
Dineshkumar Muthu
Flight Specialist | Tier 4
Dineshkumar Muthu
Flight Specialist | Tier 4
August 6, 2026

Exploring AI + Automation Anywhere for Bot Failure Log Classification — looking for community insights

  • August 6, 2026
  • 1 reply
  • 12 views

I’m planning to integrate an Ollama LLM with Automation Anywhere using the REST Web Service action to automatically analyze bot failure logs.

The goal is to group and categorize errors such as:

  • Credential / access issues

  • Application not responding

  • Selector or object change

  • Data format mismatch

  • Environment or profile corruption

  • Network / timeout errors

  • Excel/Office activation popups

  • Unexpected UI popups

Once the logs are classified, we can automate the production support workflow, including:

  • Auto‑creating PRD support tickets only when required

  • Routing tickets to the correct support queue based on error category

  • Providing recommended fixes directly inside the ticket

  • Reducing manual triage time for support teams

I’m curious if anyone here has used AI models (Ollama, OpenAI, Claude, etc.) with Automation Anywhere for similar production support automation or log intelligence.

Would love to hear your use cases, lessons learned, and model choices.

Tagging for visibility:  @Developers @Community Captains @MVPs @ Community Leaders Your insights would be super valuable for anyone exploring AI‑driven bot reliability and automated support operations.

1 reply

Aaron.Gleason
Automation Anywhere Team
Aaron.Gleason
Automation Anywhere Team
August 6, 2026

​@Dineshkumar Muthu Most of what you are asking here is available by explaining to the LLM in the prompt what each of these errors appears in the logs. 

Advising an LLM about what the log looks like when a Recorder: Capture fails will get you in the ballpark of the issue, but you have to look closely -- do you have enough information to know if this is an unexpected UI popup or a web page that didn’t load properly… or Chrome displaying the “What’s new!” page instead of the expected page.

The LLM isn’t magic: If you can’t textually describe the problem and how to detect it, the LLM won’t be able to detect the specific error.

Now, that being said, you don’t have to use REST Web Service calls to call an Ollama LLM. You can create a custom AI model connection. Then you can use that third-party LLM with AI Skills and AI Agents too.

 

Terms & ConditionsAccessibility statement