Think about how you use most software today. To find anything, you filter. You open a menu, pick a category, add a date range, tick a box, sort a column, then scroll. The enterprise and consumer tools you know best, the online store, the CRM, the ERP, all work the same way: filter, filter, filter until the thing you want is in front of you. You are doing the system’s work by hand, one narrowing step at a time.
That interaction model is starting to fall away, and it changes how R&D software should work. The shift is from point‑and‑click navigation to an AI‑first workflow, where you state what you want and the system reaches the answer for you.
What Is Changing About How We Use Software?
Software is learning to act on intent instead of waiting for navigation. For decades the deal was that a person translated a goal into a sequence of clicks that the software understood. Now the software can take the goal directly. Instead of opening several filters to narrow a list, you type the request and the system returns the result. The clicks disappear, and so does most of the learning curve that came with them.
That last point matters more than it sounds. Every filter and menu is something a new user has to learn before they are productive. Remove the narrowing and the barrier to entry drops. A scientist who joined last week can ask the same question as someone who has used the system for years.
Why Does This Matter for R&D Data?
R&D is full of questions that are painful to reach by filtering and straightforward to reach by asking. “Show me every sulfate‑free formulation that passed our foam‑volume target.” “Which of our coatings held gloss above specification after accelerated ageing.” By hand, those are long chains of filters across screens. Stated as a request, they can be a single step, if the system can reach the answer.
That “if” is the whole game, and it is where a general‑purpose chat assistant and a purpose‑built R&D platform diverge. A general assistant can accept the sentence but has nowhere real to send it, so it guesses from documents. A platform whose data is structured and connected can take the same sentence and return the actual matching records. The interface looks similar. The result does not.
The Catch: “Just Ask” Only Works on Structured Data
It is tempting to treat “just ask” as the feature. It is not. It is the payoff of the work underneath. A natural‑language request is only as good as the data it lands on. Point it at a warehouse of unstructured files and you get a fast, confident, shallow answer. Point it at a structured, connected data model and you get the real one.
This is the same principle that governs everything else in R&D AI: structure first, intelligence second. The reason to build AI‑first workflows is not that typing is nicer than clicking. It is that when the data is structured, stating intent becomes the fastest, lowest‑friction way to reach real answers, and the manual filtering that used to gate every question simply disappears.
What This Looks Like in Practice
Inside Uncountable, this is how the work is meant to feel. Bodie, the platform’s built‑in assistant, is designed to act on a request grounded in the project’s structured data rather than just answer trivia. A scientist can describe what they want and have the platform do it. Bodie is available with the Gen AI module, so talk to your account manager about turning it on.
The point is not the novelty of talking to software. It is that the filtering‑heavy way of working was always a workaround for systems that could not understand intent, and that workaround is ending.
The teams that benefit will be the ones whose data is structured enough to make “just ask” return real answers. For everyone else, removing the filters only removes the last thing holding a shallow system together.
Want to see “just ask” on real structured R&D data?: Request a demo
.png)
.png)
