On day 1 every position from CEO, CFO, to mail delivery messenger is filed with one or 2 names: the founder(s). As the company grows, you hire people and start delegating the work so that they can fill the positions on the organization chart.
Mulling over an org chart is one of the easiest of the thousands of nice-sounding things you need to brutally cull from your todo list in the early days.
So, early org chart is only useful if it doesn't jump too far ahead of what the business is: If you discover you need a different organization to execute on the plan, there's no point in planning for scaling the existing one out. But if what's being followed is a "hire a great team and great things will happen" kind of plan, then there's no conflict since it just means you've considered both near and long-term aspects of building a great team.
The GP is clearly suggesting thinking about role definition from day one rather than waiting until the 10-20 employee phase. I'm stating unequivocally that I think that is a mistake.
The "urgency of hiring and delegating" only kicks in once you have a level of certainty of what needs to be done. Every employee you bring on adds not only burn rate, but also communication overhead. The only thing that gets a startup from the garage phase to actually being able to series A and credibly hire more than 10 people is for every founder and early employee to ruthlessly identify things that need to be done and execute on them. This idea of documenting what exactly founders are doing and thinking about delegating them is what YC has dubbed "playing house"*.