As the application layer of AI takes off, the big question is bubbling of what happens to our aggregated data that becomes our personal source of truth. Who owns it? Who controls it? How is it accessed, stored, and protected? At the individual level, we think of this as memory. But at the organizational level, we think about this collective memory as a system of record. 

Throughout software history, becoming a system of record has been the key to move from tool to platform and go from utility to dependency. Salesforce didn’t just help with CRM, it became the source of truth for customer relationships. Carta did the same for cap tables. Github, for code. These companies broke out and sustained value by becoming the places where knowledge lives.

Sometimes, this emerges over time. Figma, for instance, began as collaborative design software. But its breakout power came when it evolved into the canonical source of product design specs and assets. A record not just of work done, but of how decisions were made.

Today, we’re witnessing a total rebuild of the tooling layer. AI is eating manual workflows. Automation is trivial and data flows in real time. The velocity is addictive. But it also raises the nagging question of if tools are easier than ever to build, integrate, and replace, what creates stickiness?

The answer likely isn’t faster features. It’s shared memory.

Tools drive efficiency. Systems of record build dependency, and ultimately, longer-term moats.

There is a large and interesting wave of new tools streamlining healthcare admin. Automating outbound calls to retrieve medical records, for example, is a clear efficiency win. But the real unlock is storing those records in a persistent repository. The next time a patient calls, the data is already there. That’s a system of record.

Recording meetings and automagically generating perfect notes has been a major efficiency unlock. Granola is my most addictive workflow product. But when those notes are automatically tagged, shared, and searchable across an org we will build a living, collective memory of every company conversation. That’s a zero-input CRM. That’s a system of record.

In legal tech, AI tools help lawyers draft and analyze with speed. But the next step—storing those drafts, comparing new terms to redlined standards, flagging deviations—that’s the shift from utility to platform.

We’re excited by tools that retrieve, move, and process data with elegance. But we’re interested in partnering with tools that turn this data into collective knowledge and sources of truth. Despite a wildly changing landscape, systems of record continue to be the door to endurance. 

As the application layer of AI takes off, the big question is bubbling of what happens to our aggregated data that becomes our personal source of truth. Who owns it? Who controls it? How is it accessed, stored, and protected? At the individual level, we think of this as memory. But at the organizational level, we think about this collective memory as a system of record. 

Throughout software history, becoming a system of record has been the key to move from tool to platform and go from utility to dependency. Salesforce didn’t just help with CRM, it became the source of truth for customer relationships. Carta did the same for cap tables. Github, for code. These companies broke out and sustained value by becoming the places where knowledge lives.

Sometimes, this emerges over time. Figma, for instance, began as collaborative design software. But its breakout power came when it evolved into the canonical source of product design specs and assets. A record not just of work done, but of how decisions were made.

Today, we’re witnessing a total rebuild of the tooling layer. AI is eating manual workflows. Automation is trivial and data flows in real time. The velocity is addictive. But it also raises the nagging question of if tools are easier than ever to build, integrate, and replace, what creates stickiness?

The answer likely isn’t faster features. It’s shared memory.

Tools drive efficiency. Systems of record build dependency, and ultimately, longer-term moats.

There is a large and interesting wave of new tools streamlining healthcare admin. Automating outbound calls to retrieve medical records, for example, is a clear efficiency win. But the real unlock is storing those records in a persistent repository. The next time a patient calls, the data is already there. That’s a system of record.

Recording meetings and automagically generating perfect notes has been a major efficiency unlock. Granola is my most addictive workflow product. But when those notes are automatically tagged, shared, and searchable across an org we will build a living, collective memory of every company conversation. That’s a zero-input CRM. That’s a system of record.

In legal tech, AI tools help lawyers draft and analyze with speed. But the next step—storing those drafts, comparing new terms to redlined standards, flagging deviations—that’s the shift from utility to platform.

We’re excited by tools that retrieve, move, and process data with elegance. But we’re interested in partnering with tools that turn this data into collective knowledge and sources of truth. Despite a wildly changing landscape, systems of record continue to be the door to endurance.