There is something I have noticed after years of working with founders, executives, and teams: most companies are far better at doing work than they are at learning from the work they have already done. Every week brings new meetings, decisions, customers, problems, experiments, and lessons, but much of that intelligence disappears almost as quickly as it arrives. A salesperson discovers what finally moved a difficult deal forward, a manager figures out how to handle a difficult situation, or a product team learns why something did not work. Everyone moves on, and a few weeks later someone is forced to learn the same lesson again.
That is not a people problem. It is a systems problem. We have built incredible systems for storing information, but very few for building organizational memory. Slack stores conversations, Notion stores documents, Salesforce stores customer activity, project management tools store tasks, and HR systems store employee records. Each captures a piece of what happened, but very few understand what the team actually learned from it or how that learning should influence what happens next. A company can have years of data and still have the same conversation every Monday because nobody has turned that information into usable context.
The best teams I have worked with develop a kind of collective memory. Someone remembers that a particular customer responds badly to a certain pitch. A sales leader knows that a specific objection usually means something deeper is wrong in the deal. A product manager remembers why a feature was deliberately not built six months ago. A founder knows that someone performs exceptionally well when given ownership but struggles when every decision is second guessed. These things rarely appear in a dashboard, but they influence important decisions every day. The problem is that this knowledge usually lives inside individuals, and when those people leave, change teams, or become too busy, part of the company’s intelligence goes with them.
I wrote about the first part of this problem in The Most Expensive Conversation Is the One That Never Happens. Some of the most important signals inside an organization exist inside conversations, yet most companies have no real way to capture the understanding created through them. Even when the right conversation happens, the insight often disappears because there is no mechanism for turning it into institutional memory.
This is where I think the next generation of AI becomes much more interesting. We have spent the first wave of enterprise AI asking it to write emails, summarize meetings, create presentations, analyze spreadsheets, and draft proposals. Those applications are useful, but they remain largely transactional. The much bigger opportunity is for AI to understand the context in which work happens and help the organization learn from it.
Imagine a system that does not simply know what your team discussed on Monday, but understands what was decided, why it was decided, what assumptions were behind it, and what happened afterward. The next time a similar situation appears, the team should not have to start from zero. It should already have the context from the last time and be able to use that experience to make a better decision.
This becomes increasingly important as companies grow. A ten person company can communicate largely through direct relationships because everyone knows what everyone else is working on. As the company reaches fifty or one hundred people, information becomes fragmented, meetings multiply, functions become specialized, and the founder can no longer be part of every conversation. People ask questions someone else has already answered, teams repeat mistakes another team has already experienced, and decisions are made without the context behind previous decisions.
This is often described as a communication problem. I think it is more accurately a context problem. The company has the information; it simply cannot move the right context to the right person at the right moment.
The team should not have to rely on one person remembering everything, and the founder should not have to sit in every meeting. The system should learn alongside the team. What worked should become part of the playbook. What failed should become a lesson. Important decisions should retain their reasoning. Customer feedback should influence future conversations. Individual strengths should become more visible, while managers get better signals about where someone needs support before a problem becomes obvious.
Most management systems are backward looking. They tell you what happened last quarter, whether someone hit a target, or how many projects were completed. But great teams are constantly changing, so the more important question is whether the team became better because of what happened this week.
A sales team that learns from every lost deal can improve its approach with every conversation. A product team that remembers why previous decisions were made can avoid revisiting old debates. A leadership team that understands how its people are evolving can make better decisions about responsibility and development. This is what compounding looks like inside an organization: learning from the work and making the next piece of work better.
I have always believed that one of the biggest advantages of a great organization is not having the smartest person in every room. It is making the people inside the organization smarter over time. A great team gives people context they did not have yesterday, exposes them to lessons they would not have discovered themselves, and turns individual experience into collective knowledge.
AI has the potential to make that possible at a scale we have never had before. The goal should not be another tool employees have to manage. It should be something that works alongside the team, remembers what matters, surfaces what is relevant, and helps people make better decisions without creating another layer of administration.
In many ways, I think this is where the traditional Chief of Staff concept is heading, even if we never call it that. The best Chiefs of Staff create context, remember what matters, connect conversations that would otherwise remain disconnected, and understand what is happening across an organization well enough to bring the right information to the right person before they have to ask. Historically, only a small number of companies could afford that capability. AI changes the economics.
A team of ten should be able to have the same organizational memory advantage that once required a senior operator sitting beside the founder. A growing company should not lose its context simply because there are too many conversations for one person to follow, and employees should be able to benefit from the collective learning of the organization rather than discovering everything independently.
The companies that figure this out will have an advantage that may not appear as a new feature or piece of technology. It will show up in faster decisions, better execution, stronger development, and fewer repeated mistakes. One customer insight improves the next conversation. One coaching moment changes the next challenge. One lesson becomes part of the team’s institutional memory. Over time, the organization is no longer simply accumulating information. It is accumulating judgment.
That is when a team becomes something very different: a team that gets smarter every week. And I believe that may become one of the most important competitive advantages a company can build in the AI era.
