Start by finding them. Take a task you would hand over and write down how you do it, step by step. Somewhere in that list you will hit a point where the honest instruction is that it depends, and that is the thing worth attention. It depends on what, exactly? What would you look at? What would make you choose differently? Answering that turns tacit knowledge into a rule somebody else can apply, and it is the difference between a process that works when you are not there and one that generates a phone call.
Sequence the work by frequency rather than by importance. The task you perform daily is worth documenting before the quarterly one, even if the quarterly one feels weightier, because the return compounds. And write it while doing the task rather than from memory, since the version you remember has the awkward parts smoothed out and the version you perform contains them.
Expect the first handover to produce worse results than you would achieve yourself, and plan for it rather than treating it as evidence the delegation failed. Somebody new following a written process will be slower and will make different choices. The question is not whether the output matches yours exactly but whether it is good enough and improving. Taking a task back after one imperfect attempt teaches everybody that delegation is theatre, and it is the most common way this transition stalls.
The deeper shift is what you stop doing. A founder who documents everything and still approves every decision has built a system that requires them at a different point rather than one that runs without them. Decide in advance which decisions genuinely need you, which need to be told to you afterward, and which do not need to reach you at all. That list is uncomfortable to write and it is the actual transition. Everything else is preparation for it.
There is a practical test for whether any of this is working, and it does not involve reviewing documentation. Take a week away, or at minimum a stretch of days where you are genuinely unreachable, and see what breaks. The things that broke are the things still living in one head, and the list they produce is more honest than any audit you would conduct while present. Businesses that never run that test discover the answer at a moment they did not choose, which is usually an illness or a family obligation rather than a holiday.