Features stopped starting with me
Every big feature started with me on a client call. Recording the calls and handing the definition to the team felt uncomfortable — until they stopped needing me.
Julio, is the PRD ready? For years, that question found me every single week.
And the problem was never the PRD. The problem was that the answer depended on me. I was the knowledge keeper: I joined the client calls, understood the problem, defined how the product should work, and wrote the PRD. Every big feature started with me.
The cost nobody puts on a board
The team had no ownership, because I was the bridge between the customer and the product. A bridge feels useful — you're in the middle, everything flows through you, everyone needs you. That's exactly what makes it expensive. Nothing starts until you're available.
When I came back to the operation at Riqra, that had to change. I couldn't ask a team to build autonomously if the starting point of every feature still passed through me.
What I actually did
Something almost embarrassingly simple. I started recording the client meetings, and I asked the team to do what I used to do: define the product from the call and the available context, and write the PRD themselves.
The first ones were uncomfortable. Definitions I would have written differently, edge cases I would have caught earlier. I reviewed, corrected, and resisted the urge to take it back.
It was uncomfortable. Then it stopped needing me.
Today the team defines the product and writes the PRD. I'm out of the loop. The customer's voice goes straight to the people building — it doesn't pass through my summary of it.
Why this comes before any AI tool
This is the part most teams skip. Having AI and agents changes nothing while the knowledge is trapped in one person. Agents multiply whoever holds the context — if that's only you, you've automated your own bottleneck.
First you remove the bottleneck. Then the team can multiply.
So the question is: how many things on your team still only start when you're available?