A letter from the founders
Why does PersonByte exist?
Knowledge isn't in the documents
Documents capture the end result: usually, what was decided. They almost never explain why the decision was made or how people got there.
The why lives in people: the signals they picked up on, who they talked to, the alternatives they ruled out, and the cases where they would have decided the opposite. That knowledge takes years to build, and it can walk out the door with a single resignation letter.
What makes an organization unique is the way its people make decisions. Its manuals and documents tend to look like everyone else's.
AI averages
Artificial intelligence is very good at producing plausible answers from existing patterns. Those answers tend to converge on what's most common, and today anyone can get them.
Expert judgment is precisely knowing when the case in front of you isn't the common one, or calls for a different approach. An organization that hands all its decisions to a system that averages will end up deciding like the average, and it is bound to disappear.
The opposite risk is just as real. An organization that doesn't use artificial intelligence and has no record of what it knows will disappear too: it will be slower than everyone else and always one resignation away from losing what makes it different.
Both futures worry us, and we don't believe either one is inevitable. The question isn't whether to use artificial intelligence, but what belongs to it and what belongs to people.
What we don't know yet
What makes us irreplaceable as human beings at work? How many of our decisions can a machine make today, and how many will it be able to make tomorrow? We don't have those answers, and we're wary of anyone who claims to.
We don't believe people are always right, either. Sometimes deciding differently is a sign of judgment; other times it's just the difference between a good day and a bad one.
What we do believe is that this difference can be recorded and observed case by case, instead of debated in the abstract.
Our bet
The machine applies the recorded knowledge. The person questions it and changes it.
What a person has already solved a hundred times can be available to the whole team, at any hour, anywhere. That frees them up for case one hundred and one: the one that isn't in any manual or any document.
When the culture says A, a person may see something no rule anticipated and decide B. If they can explain why, that decision can become tomorrow's standard.
How we work
- We ask; we don't just read. We start with the documents and follow up with interviews about real decisions, because knowledge is rarely written down.
- Every piece of knowledge has a clear author. The person who built it reviews and approves it. Nothing new gets recorded without their sign-off.
- When the assistant doesn't know, it says so. If it doesn't have enough to go on, it escalates the question to a person.
- We record every B. When someone decides differently from what was recorded, we document the case and the why. Over time, this shows how much of the work follows rules and how much still depends on living knowledge.
Why we're called PersonByte
Physicist César Hidalgo uses the term personbyte for the maximum amount of knowledge and know-how a single person can accumulate. Anything that requires more than that, which is almost everything worth doing, has to be spread across many people. That's why what an organization can achieve depends on how well knowledge flows among them.
We believe that limit can be moved. It's not about one person knowing more; it's about what they already know no longer depending on their time and presence. When that knowledge is recorded and available to the team, the person gets back the room to learn what nobody knows yet. That idea is where our name comes from.
Why we exist
We're not out to prove that machines can decide for people. We want to show how far people can go when repeatable work stops consuming them.
We exist so the knowledge of the people who know isn't lost, and to make it clear where they're still needed.
Roberto
Founder, PersonByte