By @Ghada Shebl and @Hassan Tayyab
AI has changed how we build product at Secfix, but not why.
The biggest shift is not that code appears faster or prototypes are cheaper to produce. It is that more iteration now happens earlier, before the work becomes expensive. We can pressure-test ideas, challenge assumptions, and get to a stronger first version before asking more people for time.
That has changed both product design and engineering.
We still care about the same things: clarity, judgment, quality, and production readiness. AI has not made those less important. It has made them more important, because the teams that benefit most are the ones that already know what good looks like.
Before AI, more of the early uncertainty lived in human review.
A rough idea would become a rough design. A rough design would go to review. A half-shaped implementation would become the place where some of the real product thinking finally happened. That process worked, but it was expensive. It consumed calendar time, engineering time, and a lot of back-and-forth that was really just the team trying to get to a clearer starting point.
Now much more of that happens earlier.
We use AI before we ask more humans for time. That lets us test a direction, rewrite it, prototype it, question it, and improve it before the work reaches design review or code review. Human review still matters just as much. The difference is that it now happens on stronger work.
A lot of product work at Secfix now starts with a conversation.
We talk through the problem with Claude, not because Claude makes the decision, but because it helps us explore directions quickly. We can pressure-test an idea, reframe it, try alternatives, and decide whether something has enough shape to keep going.
That conversation often turns into a rough prototype — something concrete enough to tell whether the idea actually holds up.
One of early Claude prototypes
Final design
After we build the interface in Figma using our components and style guide, we often bring it back into Claude for another pass.
That helps us question the UX, tighten the copy, and surface edge cases we may have missed. By the time the design reaches human review, it is usually more thought through and much easier to discuss.
That has shortened review cycles meaningfully — not because we skip review, but because the work that reaches review is already stronger.
A small but useful part of this loop is voice. Tools like Whisper Flow make it easier to walk through a design out loud instead of typing every thought. That shortens the iteration loop even more.