I am exploring the implementation of an AI-powered(LLM based) log analyzer for Robot Framework (RF) using Large Language Models (LLMs). The goal is to enable LLMs to interact with RF execution logs (HTML/XML) to automate analysis, identify patterns, and extract actionable insights. Below is my key question:
How can we best structure the pipeline for LLM-log interaction?
Example:RF Logs → Preprocessing → LLM Query Engine → Insights
I can only recommend a Framework which will allow tool calls to your LLMs. I’ve used pydantic AI in some projects.
Do not parse the xml result files directly, but use the robot api to do that.
You can e.g. store the results in a database and use a LLMs with tools to do queries to the database.
A lot of options there.