Issue 002Note from the front line
Specifying before building is leadership too
Specifying well doesn't slow a team down: it keeps early speed from being paid for later in incompatible interpretations, rework and technical decisions without context.
One of the most expensive phrases in a product and engineering team is: "we get it, let's start".
Not because it's always wrong, but because it often shows up right before each person fills in the gaps with their own mental model. Product imagines one experience, design another, backend a third, and the real conversation only happens once the code exists and changing it costs reputation, time and focus.
Why this is leadership too
Specifying isn't a paperwork step. It's a leadership intervention: organizing context, exposing assumptions, making trade-offs visible and giving the team a shared basis for deciding.
When that basis doesn't exist, the problem isn't only technical. It becomes political too. Nobody wants to stop an initiative that "has already started", even if everyone suspects it's being built on different interpretations.
Where AI does help
I find AI more useful before the code than during the code when the idea is still unstable. It helps structure PRDs, RFCs, user stories and uncomfortable questions that a busy team tends to push off until later.
That doesn't replace judgment. Quite the opposite: it forces you to review more carefully what each document is saying, what was assumed and what isn't yet worth automating. WebspecAI was born from exactly that need: compressing the time between a fuzzy hypothesis and a specification that lets you work with less noise.
What changes in practice
A team doesn't need endless specifications. It needs enough clarity to understand the problem, the scope, the constraints and the signal that would say "this worked".
When that clarity shows up early, the code gains real speed: fewer reactive discussions, less sideways rework and fewer architecture decisions made to prop up product ambiguity.
Practical tool
Five questions before writing the first line
- What concrete problem are we solving, and for whom?
- Which parts of the experience are still hypotheses and which are already constraints?
- Which technical decision would be conditioned by a product definition that is still ambiguous?
- How will we know this worked without arguing about it after launch?
- Which uncomfortable question should we answer today so we do not pay for it mid-sprint?
Connection to the ecosystem
When an idea is still blurry, racing straight into code tends to look like progress. Often it's just an expensive way of postponing the conversation that was missing.
Reply to this issueWhich product or architecture conversation is your team avoiding today, and how much rework could it save by having it this week?
Julián Engineering Manager, product builder and founder of Pambú Tech juliancardozo.github.io · LinkedIn
P.S.: If you're shaping a new initiative, I can share the minimal outline I use to turn a blurry idea into a workable specification.