Lakehouse RT, AI Gateway, Omnigent: Databricks Announcements

0:04Hey, buddy. >> How's it going, man? >> All good. >> Happy to see you. >> Yeah, it's happy to see you, too. I know we're presenting in like an hour or so. >> Yeah. >> You're going to do all the talking, so I'm looking forward to that. >> Well, we build it together, so it's going to be teamwork. >> Yeah, that's teamwork. Teamwork makes the dream work. >> Exactly. So, today, I mean, first keynote took place this morning. >> Yeah. >> Like, tell us what are your three main,

0:28like, favorite announcements from today's keynote. >> Um I would say my my favorite one was definitely definitely Lakehouse RT. So, I think that is uh I've I've known about it for a few months now, and I was very much looking forward to seeing it come out, and I got to test it

0:51beforehand, and it it seems very good. I I I I'm excited for the future for that one from uh low from uh high concurrency low latency. You don't have to use additional table tops or anything like that. It's just just works with what you have in Databricks. Just use a different compute, so I'm looking forward to that. Uh I think second, I'm looking

1:13I want to say uh customer lake. I'm looking forward to >> offering. >> Yeah. I haven't really spent a lot of time understanding it yet or what the offering is going to be or is not going to be, but that is one thing that I'm like very much intrigued in understanding where that's going to go in, and um I don't know if so much of us as an announcement today, but I I'm just

1:35looking forward to seeing seeing uh AI gateway being used more and more uh for making sure that people understand where their AI spend is going, how everything is being governed, and uh so, yeah. So, I I love that. >> That's looking more like AI gateway will be way easy to use. Uh basically, we'll just go select uh AI

1:57gateway. You can tag your models and you can specify the shared budget. For example, I don't know, your team, you can give to your team five $5,000. >> Yeah. >> And then you can give to every specific user a budget of $100. Once the shared budget is reached, you have two options, either send an alerts or block the usage.

2:20>> Yeah. >> And you can do the same thing for all the models and Unity Catalog as well. >> Yeah. You know, I I think like I have many conversations with business leaders today and the concept of AI spend was definitely top of mind for most of them. Like I heard so many scary stories from a lot of different people, not so much on Databricks, but just you know, working directly with different models and so I think that's

2:43AI Gateway tackles a very real business need, especially in an era where everyone is saying token maxing and I'm like, well, let's do ROI maxing value maxing. I think Patrick Wendell said that yesterday at the product advisory board, so yeah. >> Yeah. And what about Omnigents? >> Uh I'm looking Technically, that was not a data Databricks announcement, so but yes, I'm

3:06looking forward to spending more time on that. Got to spend like an hour when it was released. I got to see a little bit of the preview. And honestly, it's it depending I'm looking forward to actually testing out to see what works and what doesn't work as well right now, but it has the potential to be a tool that I'm going to use it for most of my regular everyday chat ChatGPT

3:29cloud type usage, so. >> Perfect. Thank you very >> In Lake House, rising to seize, Lake flows streams the flow. >> [singing] >> Iceberg and Delta Lake, tables a glow. Unity Catalog governs Lake House OLTP strong. Genie text to SQL MOSAIC AND KEAP PROLONGS. YOU sipping [music] quicken co-hosts like the fire. Data dreams scale higher and higher. Next

3:53[music] gen drop it.