Management is a memory problem. Most organisations treat it as a scheduling one.
Meetings have become the primary venue for decision-making, coordination, and consensus-building. This happened gradually, and for understandable reasons: distributed teams, cross-functional dependencies, the need to synchronise people working in parallel. The meeting solves for the moment of contact.
A meeting is a synchronous event. When it ends, its content migrates into the memories of the people who attended. The reasoning that produced a decision — the tradeoffs weighed, the constraints that were live in the room, the concerns that almost changed the outcome — survives as fragments. Those fragments decay and diverge at different rates in different people.
The result is a pattern that runs through most of organisational life. Every one-on-one, every retrospective, every planning session starts not from an accurate record of what was last decided and why, but from whatever survived. The room had the context. The next room has to reconstruct it.
Organisations have built excellent infrastructure for scheduling work, communicating about work, and tracking the output of work. What they have built almost nothing for is retaining the context of their own decisions. The knowledge a team accumulates through its working life — what was decided, why, what commitments were made, what should inform what comes next — has no home.
For a manager running dozens of one-to-one relationships, the overhead of that reconstruction is real and continuous. Every one-on-one requires reassembling the thread from wherever it was last left. Every conversation begins from whatever fragments of the previous one survived. This is treated as normal friction. The loss is quiet and it compounds.
COSTA is built on a single premise: that this is an infrastructure problem, and that it is solvable.
The design position that follows from that premise is opinionated: every significant interaction in a working day is structured data, and treating it as such is the precondition for working more intelligently with it.
One-to-one notes are the primary record of every working relationship. Written directly, dictated, or captured from a transcript, they accumulate into something that no single session makes visible. COSTA reads them before every synthesis involving that person. Across enough sessions they stop being a record of individual conversations and become a picture of the relationship.
Transcripts enter the system as source material — synced from Granola, or added directly for other tools. Their job is to be queried: for the commitment given and then forgotten, for the decision that was unmade without anyone updating the original. When the transcript is in the system, the meeting is no longer ephemeral. Tasks surface from notes and transcripts rather than requiring separate entry. A commitment made in a one-on-one becomes a task because it was already written down.
Everything is stored as plain Markdown files on the user's machine. No proprietary format, no lock-in. The value of the system is in what it accumulates. Ensuring that data outlives the application is a design principle.
Every working day produces context. The question is whether it accumulates or disappears.
COSTA produces a briefing at the start of each working day, updated throughout working hours. It reads the one-to-one notes before writing, so it knows what is open in each relationship without being asked. It reads the task list, so it knows what was promised and to whom. It reads the calendar and flags which of today's meetings continues an existing thread and which requires a position to be formed before the room assembles. More captured, more possible.
The return is deferred and invisible until it isn't. It materialises as the meeting preparation already written from notes taken last week. It materialises as the task extracted from a conversation you barely remembered having. It materialises as something harder to account for: a pattern across six months of one-to-one notes, not visible in any single session, about what a person is carrying, what tends to stay unresolved, how they respond when the work gets hard. That is not a contact record. That is an understanding.
This is why COSTA includes a daily journal. Operational records capture what was decided and what was promised. The journal captures how it felt to do the work. Very few people have a reliable picture of how their energy has tracked across a quarter, or whether the difficult week was an outlier or a pattern. That information is lived every day. When it is in the system, it can inform what gets prioritised — a run of low-energy entries is context, not noise.
The working life generates its own record continuously. Most of it disappears. What a manager retains is not a matter of memory. It is a matter of infrastructure.