After seeing how brittle and time-consuming rule-based log parsing can be in real-world environments, I set out to build LogSentinelAI, an open-source, LLM-powered log analysis tool. LLM-powered log analysis tool. Whereas traditional solutions force you to normalize logs and write rigid parsing rules, LogSentinelAI works on both structured and unstructured logs: you simply define the fields you care about, and the LLM automatically extracts, semantically analyzes, and understands context across log entries. In our Demo, we even showcase visualization with Elastic Stack.
I’d love feedback from system/security engineers, developers, and ops folks:
How accurate is field extraction on your real-world logs?
Does performance (throughput, latency) and cost feel reasonable?
Are there any security or privacy concerns I haven’t considered?
After seeing how brittle and time-consuming rule-based log parsing can be in real-world environments, I set out to build LogSentinelAI, an open-source, LLM-powered log analysis tool. LLM-powered log analysis tool. Whereas traditional solutions force you to normalize logs and write rigid parsing rules, LogSentinelAI works on both structured and unstructured logs: you simply define the fields you care about, and the LLM automatically extracts, semantically analyzes, and understands context across log entries. In our Demo, we even showcase visualization with Elastic Stack.
I’d love feedback from system/security engineers, developers, and ops folks:
How accurate is field extraction on your real-world logs?
Does performance (throughput, latency) and cost feel reasonable?
Are there any security or privacy concerns I haven’t considered?
Your honest critiques and improvement ideas will be hugely valuable. Full details and examples on Medium: https://medium.com/@call518/logsentinelai-practical-llm-base...