Ask your questions for Kelsey <!here>! I’ll make sure he gets them👇🏻

Q: AI within Platform Engineering sometimes come across as snake oil. Some miracle cure that will make us all obsolete. What do you think this AI hype will end up with in terms of system administration at the tail end of the hype cycle?

where do you think kubernetes will be in 5,10,20 years

I’ll be a conservative “old man yells at cloud” type - but.. AI is a hype at the moment. Building trust in its produce will take a significant amount of time that will potentially stretch past our lifetime. Pattern matching at scale is surely helpful, calling it “innovation” is something I’m struggling a lot with. Innovation is something noone’s heard of, often contradicting all existing global knowledge. Digesting information en-mass at scale is not innovation per se.

is backstage key to platform engineering?

“is backstage key to platform engineering?” Think a lot of people have strong answers for that one @Xafah 🤣

Still no answer!

If you care for other people’s opinion - I see it as an enabler more than a “key”. Depends on what you define as “key” 🙂

I don’t really think it’s key. I think there are lots of things you could use instead of it. Ultimately, I would say you need to meet developers where they want to work and build/buy/use what works for your team(s). So you could have a platform interface that has no gui at all and only a CLI and REST API interface b/c that’s what your devs (and other stakeholders) want. You could just have all your setup in a config file or whatever they check in with the code. Also you could make your own GUI and forego backstage. Or you could purchase something off the shelf like Humanitec, Opslevel, or Port. Just depends on your needs and what you can dev, pay for, and/or maintain.

A GUI is not necessary when building an IDP. IDPs are Products and should be approached with a PM hat on. Identify and cluster sort big buckets personas. Interview those personas so you can see what problems they have. You should probably start with identifying bottlenecks in each personas flow. If the bottleneck is solved with a UI, go for it! If not, don't build one. Focus on building the thing that removes the bottleneck. Theory of Constraints is valid here and should be used IMO.