It’s The End Of Observability As We Know It (And I Feel Fine)
It’s The End Of Observability As We Know It (And I Feel Fine)
The history of observability tools has been about making terabytes of heterogeneous telemetry data comprehensible to human beings. AI is changing this paradigm, potentially commoditizing analysis and instrumentation. The key to success is fast feedback loops, collaborative knowledge sharing, and treating everything as an experiment.
Note
a bit hyperbolic (depending on how you want to define "end" and "as we know it") but the message highlights something worth boiling down: tight feedback loops are great, keep making them tighter.
LLMs deployed as summarizes, synthesizers and translators do a great job of leveraging their strengths (assuming generating the cost of generating them doesn't outweigh the benefits).
tighter feedback loops don't just apply to telemetry either. having an LLM be a first line code reviewer also leans into another feedback loop that's worth pulling down.
highlighting the human role in these feedback loops also gives us greater clarity and alignment for the value these tools has to offer, while avoiding the hype of AI.
Published
Summary
Image
