I walked into Jio World Convention Centre on June 17 expecting the usual mix of swag, keynotes, and hallway small talk. Two days later, what stuck with me wasn't any single talk. It was a shift in the conversation itself: cloud native is moving past "how do we observe everything" and into "how do we get something to act on what we observe."
I was there for KubeAutoDay on June 17 and KubeCon + CloudNativeCon India on June 18 and 19, representing DrDroid at the booth. Here's what stood out, where I think things are headed, and what's worth looking into next.
KubeAutoDay: the warm-up nobody expected
KubeAutoDay ran for the first time in India this year, the evening before KubeCon proper started. It felt less like a standalone conference and more like the pre-party: smaller room, more intimate conversations, people comparing notes before the main event got loud. We sponsored it the same way we sponsored KubeCon, but the tone was different. Less pitch, more "okay, what's everyone actually building?"
What we saw: two days of booth conversations
Most of my two days at KubeCon were spent doing two things: getting the booth ready before doors opened, and then standing at it, explaining what DrDroid does to whoever walked up.
That second part is where the real signal was. I wasn't sitting in sessions hearing what speakers think comes next. I was hearing it straight from engineers and platform leads who stopped by, asked what we do, and then told me what they're actually trying to solve right now. A lot of those conversations circled back to the same place: people aren't asking "how do we monitor more," they're asking "how do we get something to act on what we're already monitoring." Automation and AI-driven response came up constantly, not as a buzzword, but as the next thing teams are actively trying to figure out how to adopt.
Where we're going
That's exactly the bet DrDroid is making. We're not trying to give teams another dashboard. We're building toward incident response that doesn't wait for a human to start digging: auto-investigation, service discovery, an MCP server that lets agents reason across incident data. The conversations at the booth this year confirmed it. Almost everyone who stopped by wasn't asking if this direction makes sense. They were asking how soon they could get it.
Good topics to talk about next
A few things came up enough times at the booth that they're worth writing about properly:
What teams actually mean when they say "AI agents for incident response." The phrase gets thrown around a lot. The expectations behind it vary more than people admit.
Where automation should stop and a person should take over. Every conversation eventually got here, and nobody has a clean answer yet.
What adopting OpenTelemetry looks like for a team that's mid-migration, not greenfield. Most people we talked to are somewhere in the middle, not starting from scratch.
What changes for a platform team once incident response is built in from day one, instead of added after the first bad outage.
Find us there next time
If you stopped by the booth this year, thank you for the conversation. If you didn't, we'll be back at the next one. Come find us, skip the small talk, and tell us about your last incident. That's the conversation we actually want to have.
