On a Monday afternoon a few weeks ago, a close friend asked me for help. He was raising money for a real estate startup, and he wanted to talk to someone who had done it: figure out how they raised, who they went to, and what they would do differently.
I know a lot of people. That was the problem.
When someone asks “who do you know who…,” you (like me) don’t search your network. You likely search your week, the dominant part of your short term memory. The name that comes up first is whoever you talked to last, whoever emailed you yesterday, whoever happens to be near the top of your mind. Recency feels like good judgment, but it isn’t.
A few days earlier I had finished setting up a system I now use more than any other. Every person I know well has one or two sentences next to their name, in my own words, about what I would actually go to them for. As President of a Chapter in SCN called Automatic Wins, I meet several people a week, who want to be remembered.
So, once a week a scheduled run reads my inbox, without sending, moving or deleting anything, and records how recently and how often we talk. That gives every person a warmth score.
Using this skill, I typed the question into Claude the way my friend had asked it.
There was one strong match. A business school classmate who had raised money for his own real estate company. My own note on him said he was “the right person to go to for all things related to startup fundraising and real estate tech.”
He ranked 117th out of 122 on warmth. No email history on file. Our last message was on LinkedIn, in the summer of 2023.
The people near the top of my warmth list were there too, as partial matches. One was an investor who had written to me that week. One had built and sold a company. Both were people I talk to all the time, and both were the wrong fit for this question. The answer said so. For one of them it added that he already owed me two answers, so this was not the moment for a third ask.
If I had answered from memory, I would have handed my friend one of the warm names.
The right person was number 117.
The system identified the optimal friend to introduce from my own notes, not my memory, and asked me to confirm before going further.
It drafted a 90-word introduction, then suggested I text him first, which I did to a warm response. I sent the introduction that night.
And at the bottom, one line: nobody in my network who invests in real estate. That gap turned into a bigger project I’m still working on.
Every platform that holds your contacts is very good at counting them. Connections, followers, contacts synced. None of them ask you to write one sentence about why a person matters. A count is easy to show you and easy to sell. A sentence in your own words belongs to you, in the future, when you need it most.
So the network sits there, large and silent, and when the question comes you answer it from your short term memory.
This is not a new problem. Before contact apps, communities kept this memory in people. In West Africa that person was the griot: the one who knew the lineages, who was connected to whom, whose family could be trusted with what. When you needed something, you didn’t search. You asked the one who remembered.
I named my own system GRIOT for that reason. The tools changed. The job didn’t. Someone has to hold the memory, and if it lives only in your head, it is only as good as your week.
If you lead anything, people will keep asking you who you know. You can answer from memory and hand them whoever you saw last. Or you can spend fifteen minutes this week writing one honest sentence about ten people you would actually call.
The sentence is the whole system. Everything else reads it back to you.
The full setup, with a two-minute demo on invented people: https://github.com/WaziriG/skill-drops/tree/main/01-relationship-memory
The fifteen-minute version, no inbox needed: https://github.com/WaziriG/skill-drops/tree/main/02-who-knows
— Waz