The old way is dead. This is why I built Querio.
I’m going to try to not make this sound like a giant advertisement for Querio, but this blog explains the reason I believe a product like Querio exists, and why I started it.
I love independence, and achieving my potential. When I worked in product at Amazon, my favorite part was that I could discover my own project, fight for my own funding, and just do things. How did I do that? Through SQL and Python. I had top 1% permissions for databases for my role / level, and I so intimately knew the relationship between Amazon’s global supply chain and the data structure underneath it that supported it.
This meant I could pull up historical data, find low-hanging fruit, code my own pilot, then statistacally prove that my pilot had a benefit. When the numbers don’t lie, your project is approved. This was my superpower. I ended up with about $140MM in finance validated savings for Amazon’s supply chain in 3 years.
The other product managers in my teams just couldn’t do the same thing (mostly). Some of them knew some SQL, none knew Python, and most didn’t have access to the right databases. This meant they were caught waiting 4-5 weeks for a centralized data team to help them with their request (it was a mess).
This leads to a lot of issues. It leads to teams working from their gut, it leads to teams picking and choosing what questions to ask, it leads to low data literacy within an organization, and most importantly it ruins the self-esteem, impact, and independence of strong minded teams to deliver on their value.
Then, GPT3 came out, and I had an idea: What if I could take a few of our key table schema’s, and have GPT3 write the SQL to kick start my team’s SQL work (it was the part they got the most stuck on), and wouldn’t you know it: It was their favorite tool beyond any of the million’s of dollars worth of BI tools we had. (Yes Amazon bought things like Tableau licenses).
Here’s the thing with the last 25 years of BI tools: Every 5 years a new tool comes out that makes it even faster for the data team to serve the business team. That fundamental relationship of a data team serving the business team, has never been changed. Now we have LLM’s, and they can write code, and they can follow instructions.
I thought to myself, what if we did flip this on it’s head? Everyone has been promising democratized data analytics or whatever for years, but THIS this was never possible. What if the next big data tool was built for an entire organization to use. Why the heck do we have creator licenses still? I’m under the strong opinion that the best “AI” companies are building products that were impossible before AI and get better with every new model, like us.
This is the fundamental change that I think is here with LLM’s. For the first time, AI tools can act as a true ‘first line of defense’ for the ENTIRE company to get their own answers first. Not everything needs to be a ticket, people don’t need to choose what questions to ask because they can’t everything.
This is going to have huge changes that are underrated. First of all not everything is a dashboard anymore, if the marginal cost of a query is close to 0, you can write a handful of prompts instead of perfecting a dashboard. Second, SO many more insights will be uncovered. Everyone get’s to ask ALL their questions, not matter how small. Third and most important, you have 10x’d the agency of your company overnight.
This is the new reality and one we should change aggressively. We should also chase it AMBITIOUSLY! This means we give an AI context, we govern the tool, and we allow people to ask ANYTHING. This means we do the hard work of recreating a tool from the ground up, to support a world where both business and data teams will do data work. Many old BI tools (Looker, Tableau) or even databases (snowflake) are doing the same mistake of the last 25 years and using LLMs to write SQL for DATA TEAMS!
That is not the point. This companies are thinking too small. I refuse to let this be another wave of improving the way data teams serve business teams. This is the purpose of Querio. It’s not trying to help only data teams move faster, it is re-imagining data analytics software that faces the reality head on: Everyone at a company will be able to have data at their finger tips. That future is already here. We will not let these legacy companies hang on to their old ways of doing things. It use to be the best way, but it is not anymore.
If you would like to fundamentally change the way your organization iteracts with data, and empower your team. please reach out to me personally at rami@querio.ai.
