Company · 1 min read
Why we start every product under the root sign
A radicand is the number you have to work through before you get an answer. Here's how that idea shapes the AI products we build.
Radicand Labs, Founding team
In maths, the radicand is the number under the root sign. You can’t get an answer until you’ve worked through it. We named the company after it because that’s where we like to start: under the surface, with the problem itself.
Models are not products
It has never been easier to wrap a model in an interface and call it a product. Most of those products start with the model and go looking for a problem. We do it the other way round. Every product we build starts with a problem we’ve felt ourselves, in our own work, more than once.
Write the problem down
Before anyone writes a prompt, we write the problem in a sentence. For Afara it was: “Coding agents build against context they’ve never seen.” If we can’t agree on that sentence, we’re not ready to build.
Then measure
Once the problem is clear, we write the evaluation before the feature. Examples of the inputs we expect and the answers we’d accept. That suite is the spec, and nothing ships until it passes.
Then build, and use it ourselves
We use our own products every day. Afara runs on Afara. If something annoys us, it will annoy you, and we’d rather find out first.