The first piece of software we built with Constructs wasn't for a customer. It was for ourselves.
What we needed
Early on, we were doing a lot of qualification work manually — evaluating whether an opportunity was worth pursuing, whether a potential customer's workflow was a fit, whether the timing was right. It was judgment-heavy, slow, and inconsistent.
We knew what good looked like. We didn't have software that did it.
What we built
We built a qualification workflow: a sequence of steps that takes in signals about an opportunity and returns a structured recommendation with reasoning.
The workflow covered: company fit, use-case clarity, workflow complexity, and timing signals. Each step produced a structured output that fed the next. The final output was a recommendation and a set of flags — things to explore further or watch out for.
What it taught us
Building it ourselves was the best thing we could have done. We hit every hard problem our customers will hit: agreeing on what "qualified" means, deciding what signals actually matter, figuring out what to do with ambiguous cases.
The software got better as our judgment got better. That's the loop we're trying to give every team we work with.
What it looks like now
We use it on every inbound opportunity. It doesn't make the decision — it makes the conversation better. We come in knowing what we think, why we think it, and what we're still unsure about.
That's what good software does.