This is rare: an article from an AI SRE solution vendor that speaks to the risk we take in adopting this kind of tool. They share quite a bit of detail on what they are doing to try to understand and improve their tool’s accuracy.
Martha Lambert — incident.io
This was a fascinating incident! I love how Lorin shares his process for trying to understand Telstra’s incident report.
Lorin Hochstein
No summary from me, I don’t want to ruin it. Well worth a read.
Lorin Hochstein
Alex tells it like it is, and I loved it.
AI can explain it to you but it cannot understand it for you. That understanding is a key aspect of ownership.
The original version of this issue incorrectly used the term “hot take”.
Alex Ewerlöf
Error budgets are about understanding and managing risk, and the risk equation is changing with LLMs.
if a material change depends on concentrated knowledge, single points of failure, or solely on AI models, that change is riskier than a similar change the team can explain and operate without them.
Paul LaPosta — LeadDev
How do you measure kubernetes cluster utilization?
But I once had an economics professor who would respond whenever you mentioned a rate like productivity, efficiency, or utilization with a simple question: “Of what?” This was a polite way of saying: “Tell me the damn numerator and denominator, and do it right now.”
Ian O’Gorman — ACRL
A code change was deployed before the schema change it depended on.
Angelo Saraceno — Railway
Key business logic lived in the database as Groovy scripts, and analysts changed it live constantly. Why set up version control and CI/CD when everything worked fine? How they made the case is an interesting story.
Dmitry Moskalyuk
