Episode #577: The Machine Gun vs. The Sniper Rifle: How LLMs Changed Data Analytics Forever

Stewart Alsop III (0:00)

Welcome to the Crazy Wisdom Podcast. This podcast is for you if you have an insane drive to find the truth of things.

It's not the good answers that we seek, but the good questions. I interview a range of different guests from many different fields, all with the intention to uncover the simple truths that are hidden in plain sight. Most people don't want to go there. I go there. My guests go there, and you benefit.

Please let me know if you enjoy these episodes and as always subscribe on iTunes, Spotify or wherever you listen to the podcast. Welcome to the Crazy Wisdom podcast. I've got Teodor Ionita and he is the founder at Ideacon. How do you pronounce that?

Teo Ionita (0:44)

Ideaticon or Idaticon. It depends on how you want to pronounce it, but that's a good pronunciation, those two words. I'll give you that.

Stewart Alsop III (0:52)

What is it?

Teo Ionita (0:54)

It's a data consulting company. Essentially, I build data platforms for people. It's something that I've grown for the past two years specifically.

Essentially, what we do and what I do that I started as a freelancer and now I've started subcontracting and hiring people for it, salespeople, but also subcontracting to members of technical staff. I think that's how you call them. I basically build everything from ingestion up to the data warehouse and then from the data warehouse, how to consume that, how to send it to your client applications, are you going to use it in dashboards, are you going to use it in your actual chatbots, are you going to use your internal tools, operational resources, anything that you want to use the data as long as it involves a pipeline and a warehouse, I will build it for you basically.

Stewart Alsop III (1:54)

That's a pipeline and a warehouse.

Then I'm assuming that the LLMs have radically changed that workflow. Has it radically changed that workflow or is it still basic same stuff?

Teo Ionita (2:07)

Yes, 100 percent.

I think the biggest thing and the biggest, funny that you mentioned that because today Anthropic just announced that Claude is going to do dashboards for you directly.

They're going directly into this BI. The data analytics space has changed completely. Most of the data up to now was consumed in dashboards. Most of the people, that's how they would consume it up to two, three years ago, it would be dashboards or it would be reports, but those are also mostly dashboards. People would either look at some summary tables or dashboards and say, now most people ask their agents. So funny enough, SQL is highly performant to get data out in and out of places. That's what it was designed for. Actually, another funny story, Structure Create Language was designed to be the next business use case. So before LLMs existed, essentially when SQL was first designed, the idea behind it was to allow people to self-serve, like the business owners, the managers, the product managers, people were allowed, because it was very close to natural language, right? Like select star from, like it's like something very, very easy, like where like you have the filters, it was so easy for you to do any, like the joins as well, like a left join or a right join. They're all designed to be easily understandable by speakers of the English language. But obviously, that never happened. Then you had the sort of people who became SQL experts and data warehouse experts, and it became a normal programming language.

Stewart Alsop III (3:44)

Why did that not happen?

Teo Ionita (3:46)

Because it's, and it's the same thing with an, I mean, I don't want to go up there and say it's going to be the same thing with an LLM, but it's the same idea of like, you still need to, first of all, you need to understand the order of operations in SQL. You need to understand the sort of where are you going to run it, how are you going to run it, what's the most efficient way to run it. Then people just simply don't want to handle that themselves. I don't expect the manager to go themselves and write a query, even before LLMs existed, to write a query in BigQuery or in Databricks or in Snowflake or whatever, and just go and create their own reports. They're always going to outsource that to someone who will do it for a living, which is why essentially you have this business analyst or analyst or data analyst career path, which then it became... That's how it started, actually. It started around 2000... I don't even know when Snowflake was, 2010ish, 2000, 2015ish. It started with a data analyst. That was the first...