Shipping three SaaS in two years is less about heroics than it is about rhythm. Heroes burn out on month four and the product dies on month five. Rhythm survives.
Our rhythm, in four rules
Two-week cycles, always
No exceptions, no “this release is bigger so we’ll need three weeks”. If a feature doesn’t fit in two weeks, it gets split. The forcing function of a Friday demo keeps everything honest.
One product gets everyone, not everyone gets a product
When we’re on QanoonX, everyone is on QanoonX. We rotate quarterly. The counterintuitive outcome is that cross-product learnings compound instead of dying in a single PM’s head.
Ship the thing, then ship the telemetry
Every feature goes out with the metric that tells us whether it worked. We’ve killed five features by week four because the data said so. That’s fine.
Six months of compound interest
The first three products were slower to ship than the fourth will be. Because we now have a design system, a deploy pipeline, a test harness, and a set of AI eval templates that we didn’t have on day one. Boring infrastructure is where the velocity hides.
What we’d do differently
We’d hire the designer earlier. We’d write the eval suite on day one, not day ninety. We’d turn down the third client project and stay focused.
Most of “what we’d do differently” is just: be more patient earlier, so you can be faster later.