Why use Quodor Data for BI

The infrastructure for building an AI assistant is evolving extremely quickly and we’re committed to staying at the leading edge of what it’s capable of. Our bet is that a lot of the techniques for getting the most out of an AI assistant will emerge over the next few years, and only a small, nimble, adaptable company will wind up using them to the greatest effect. If you want your BI tool to launch with the best possible AI integration and remain at the leading edge, you’re better off working with a new team that’s building from the ground up and adapts quickly to change.

Our assistant is meant to imitate a helpful co-worker, sitting at your computer with you. It watches you as you use the BI tool, and it can interact with anything in the interface for you. This means that if you’re ever stuck in some tedious task, you can always ask it to finish up for you. It means that you can ask it things like “take me to the dashboard from the weekly review meeting” or “rename all the tables and columns about ‘user’ to use the word ‘client’ instead”. Unlike most other assistants that will automate just one specific task for you, ours will automate literally any interaction you might have with it.

Our assistant gets access to your sources of data and browses them in an intelligent manner. It isn’t economical for an AI assistant to keep your entire dataset in its context window, but ours will be able to hold all the important stuff it needs to know. It’ll know about outliers, trends, the possible values of a categorical field and what all that information means. This isn’t easy to do, and it involves making a lot of smart tradeoffs around when and how to query your data and feed it to the AI assistant. It’s also something that no company is very good at (yet) and we’re expecting we’ll be able to stay at the leading edge.

Your data represents something meaningful about the real world around us, and in order to work with it, you need to understand exactly what it represents. We obsess over documenting the semantics of your data. Our tool guesses and documents the semantics for you, prompts you to document new columns when they come up, and works hard to keep those docs as meaningful and as concise as possible. Compare this to the typical tool that tries to guess all the meanings from the table names and the column names. Having great semantics for your data is a huge advantage.

Quodor Data only ever allows the AI assistant to make changes to a private copy of your BI artifacts, but it allows and encourages the assistant to make big, bold changes. Your other AI assistant might help you write one SQL query or modify one chart at a time, but ours will make broad sweeping changes across the entire data pipeline, and present them to you for review so you can see exactly what it changed, how that affects your reports, and decide whether to accept the changes or to keep prompting the assistant to tweak them. Our workflow is substantially different from what most BI tools offer, it’s super easy to work with, and we think it gives you the perfect balance between using AI aggressively, and keeping your BI artifacts safe from AI errors or hallucinations.

BI tools are normally geared toward building just the visualizations of your processed data and often expect that you maintain some other data pipeline to process the data to get it ready for visualization. We firmly believe that’s not the right scope for a BI product. Our tool will be fully capable of building and maintaining the entire data pipeline leading up to the visualizations, starting from the raw data if need be. It’s important that we have that broader scope because the kinds of questions and requests you have of your AI assistant will be based on something deeper in the processing pipeline. This means that if you’re shopping for pipeline tools for the first time, you’ll be able to get 2 tools in one by using us. And if you already have a separate data pipeline tool (e.g. DBT), we’ll release an easy way to import and start managing the data pipeline within Quodor Data.

Our tool will have just about the simplest User Experience you can get away with. You learn how to use it by prompting the AI assistant, and the assistant shows you how to make the changes you want, or makes them for you. That lets you avoid having to learn the elaborate UX tools, buttons and menus and drag-and-drop interfaces that other tools force on you. This simplicity should be very useful to you: You can onboard new users very quickly, people will be able to do a lot with the tool even if they have very little data analysis or coding experience, and you’ll rarely have to look anything up in our docs.

Coincidentally, it also gives us a product that’s a lot simpler and easier to maintain than similarly functional tools, and you’ll see that reflected in our prices.