Now, let's talk about AI for BI - does your customer really want to talk to their data?

If you ask customers, I would imagine about 9/10 would say "yes," because they are imagining a magical system that is general-purpose, and could solve anything just like "ChatGPT already can" they say. Therein lies the problem—the assumption that, because the customer asked you for it, they must be willing to pay for it and use it—if only you could build it for them.

We learned something about this years ago when we offered "customized dashboards" for certain types of products. In many early analytics products, the thinking was, "blank slate, all the controls, let them build anything they want."

Except they didn't. Why?

What you've done here is offload the design and UX work to your customer; it's now their job to both frame the problem clearly, and design the solution—before they can get any value out of the tool. You gave them a blank slate, which might as well be writers' block for some.

If we just talk to customers, most will say "yes, we want customization."

But in reality, customers paying for a product often do not understand the workload that this might put on them before they can get any value out of it—if the product was sold on the merits of out-of-the-box value. (This is less true for developer/analyst tools, but not entirely). Think of it this way: they already traded something of value (money) to get access to your tool (product). Now they're being asked to make another significant effort-investment before value can be gained. That's "2 costs" now, not just 1.

Is that what they signed up for?

Frankly, I worry about simply slapping AI interfaces onto BI products and assuming this will be a competitive moat. I don't think it will be—because the goal of the customer really isn't to just "talk to the data."

The value you create is rooted in when you make a positive change in the life of the customer, the buyer, or some other KPI that is anchored to the customer's business—which is not equivalent to "use of a product feature."

How I would be thinking about AI for BI and building automated intelligence tools right now:

1. Is what the customer values really changing because AI exists? Has the pain, hassle, or challenge they need removed in the short-term feature—that thing that is worth paying money to get rid of—gone away just because GenAI exists?
2. A UX outcome is a positive change in the life of your end-user; by solving for these, we can increase meaningful adoption, reduce churn and create champions. You should be asking, "is my end user now looking for a different outcome just because AI exists now [either in your product or in their own toolkit]?"
3. What is always going to be true about your customer's world—whether or not LLMs and AI continue to exist and get better?
4. What barriers-to-entry are your competitors (or even your customers) assuming must always exist—but perhaps do not have to? (Ex: does your buyer really need to integrate directly with CRM/ERP data before your product can produce value?)
5. What is everyone else currently competing on at the cost of another attribute that a spectrum of users might find more valuable? (Ex: what if your solution was LESS ACCURATE with answers, but gave "ranges" of answers WAY FASTER?)
6. Who is a potential unserved/underserved user that is currently ignored by you and your competitors because of some existing barrier? (Ex: Claude Code removed much of the engineering barrier for Product and Designers; is it perfect code for production? Probably not; but has it created value?)

While all of the engineering and productivity capabilities of AI are significant events worth paying to and investing in, there are larger strategic questions you should be asking—many of which may fly above the noise.

Chances are, there was another model release before I got to the end of this article—but I'm fairly certain the questions above are still relevant.