Operations Agent For Triage

Let's dive into the operations agent which accelerates a large number of incident management detection based workflows. In this case, we're going to look at triage workflow that would be common to a governance team or a data operations team. Alert fatigue can be a real problem for teams in which case they're getting monitors that uh are misconfigured, have a high breach rate because they have too tight of a threshold or maybe there's no assigned audience. So either teams are getting notified too much or not at all. In this case, we can see that we have a few monitors here that are misconfigured. We can see that uh we have one monitor here with a over 90% breach rate. And here are monitors with no assigned audiences. So in this case, the monitor with a 90% breach rate, we have one monitor that's accountable for 5% of our overall alerts every single day. So this is a monitor that is not helpful and is just contributing to fatigue. These monitors are monitors that have intent. They're looking for issues but that no one is being proactively notified. So what we can do is ask our operations agent to take the next step. So for example here I'm asking the operations agent to suggest fixes for those misconfigured monitors. What is the right uh SQL that we can provide? Also the threat the monitors that are breaching all of the time. What's a better threshold for that? And then finally we're asking them to recommend audiences for orphan monitors and to explain their reasoning. This last one is pretty powerful. If there's an orphan monitor, a data governance team that's maybe removed from the day-to-day operations and has no idea who set the monitor would be spending hours collaborating across the team trying to figure out the home for this orphan monitor. In this case, the operations agent is able to not only quickly suggest a fix for the misconfigured monitor, provide a better threshold for the uh monitor in YAML, but also it's going to look at all the table names, the domains, and use all of the context clues that would take a data governance or data operations person hours to put together and make really good suggestions. Here we can see the suggested fix. Uh here we can for all of our misconfigured monitors. Here we can see uh our monitoring recommendations we have our monitor types in the subscription revenue in the finance sales domain going to a finance critical or sales recommended audience. So this was a really powerful demonstration of the operations agent and how it was able to uh identify operations that weren't quite working within the team and take the next step to recommend fixes. Very powerful functionality.