Alessio Fanelli: belief

Transcript context

…I think the way I think about it is that at the end of the day, semantic layer is a code base. And in Qubit, it's essentially a code base, right? It's not just a set of YAML files with pythons. I think code is never perfect, right? It's never going to be perfect. It will have a lot of, you know, like revisions of code. We have a version control, which helps it's easier with revisions. So I think we should treat our metrics and semantic layer as a code, right? And then collaboration is a big part of it. You know, like if there are like multiple teams that sort of have a different opinions, let them collaborate on the pull request, you know, they can discuss that, like why they think that should be calculated differently, have an open conversation about it, you know, like when everyone can just discuss it, like an open source community, right? Like you go on a GitHub and you talk about why that code is written the way it's written, right? It should be written differently. And then hopefully at some point you can come up, you know, like to some definition. Now if you still should have multiple versions, right? It's a code, right? You can still manage it. But I think the big part of that is that like, we really need to treat it as a code base. Then it makes a lot of things easier, not as spreadsheets, you know, like a hidden Excel files. The other thing is like then having the definition spread in the organization, like versus everybody trying to come up with their own thing. But yeah, I'm sure that when you talk to customers, there's people that have issues with the product and it's really like two people trying to define the same thing. One in sales that wants to look good, the other is like the finance team that wants to be conservative and they all have different definitions. How important is the natural language to people? Obviously you guys both work in modern data stack companies either now or before. There's going to be the whole wave of empowering data professionals. I think now a big part of the wave is removing the need for data professionals to always be in the loop and having non-technical folks do more of the work. Are you seeing that as a big push too with these models, like allowing everybody to interact with the data? I think it's a multidimensional question. That's an example of, you know, like where you have a lot of inside the question. In terms of examples, I think a lot of people building different, you know, like agents or chatbots. You have a company that built an internal Slack bot that sort of answers questions, you know, like based on the data in a warehouse. And then like a lot of people kind of go in and like ask that chatbot this question. Is it like a real big use case? Maybe. Is it still like a toy pet project? Maybe too right now. I think it's really hard to tell them apart at this point because there is a lot of like a hype, you know, and just people building LLM stuff because it's cool and everyone wants to build something, you know, like even at least a pet project. So that's what happened in Krizawa community as well. We see a lot of like people building a lot of cool stuff and it probably will take some time for that stuff to mature and kind of to see like what are real, the best use cases. But I think what I saw so far, one use case was building this chatbot and we have even one company that are building it as a service. So they essentially connect into Q semantic layer and then offering their like chatbot So you can do it in a web, in a slack, so it can, you know, like answer questions based on data in your semantic layer, but also see a lot of things like they're just being built in house. And there are other use cases, sort of automation, you know, like that agent checks on the data and then kind of perform some actions based, you know, like on changes in data. But other dimension of your question is like, will it replace people or not? I think, you know, like what I see so far in data specifically, you know, like a few use cases of LLM, I don't see Q being part of that use case, but it's more like a copilot for data analyst, a copilot for data engineer, where you develop something, you develop a model and it can help you to write a SQL or something like that. So you know, it can create a boilerplate SQL, and then you can edit this SQL, which is fine because you know how to edit SQL, right? So you're not going to make a mistake, but it will help you to just generate, you know, like a bunch of SQL that you write again and again, right? Like boilerplate code. So sort of a copilot use case. I think that's great. And we'll see more of it. I think every platform that is building for data engineers will have some sort of a copilot capabilities and Cubectl, we're building this copilot capabilities to help people build semantic layers easier. I think that just a baseline for every engineering product right now to have some sort of, you know, like a copilot capabilities. Then the other use case is a little bit more where Cube is being involved is like, how do we enable access to data for non-technical people through the natural language as an interface to data, right?…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.