Vol. I — The problem
The Coordination Tax
Why coordination breaks down as companies scale, where the hours and money disappear, and why adding more tools doesn't necessarily solve it.
AI context platforms are emerging
AI needs context. But what should that context actually know?
As companies move from experimenting with AI to deploying copilots and agents across the business, context is becoming part of the enterprise infrastructure.
Two short books from In Parallel explore both sides of that shift.
Why now
Context is becoming enterprise infrastructure.
Every new copilot, model and agent needs to understand the same organization — what things mean, what changed, what was decided, what matters now and how it all connects.
Gartner expects the majority of enterprises running AI agents in production to standardize on a dedicated context platform.
Gartner describes AI context platforms as an emerging market for infrastructure that constructs, governs and delivers reusable context to AI agents.
Gartner, Market Overview for AI Context Platforms, August 2026The question isn't whether AI needs context. It's what that context should know.
Our view
For us, the answer goes beyond documents and stored data.
Companies create valuable knowledge constantly — in meetings, decisions, changes in direction, commitments, risks, dependencies and the everyday work that rarely makes it into a system of record.
Without a shared understanding of that reality, people spend enormous amounts of time reconstructing it. And now AI depends on them to do the same.
Two books. Two sides of the same problem.
Vol. II — What comes next
Shared Context
What changes when the knowledge created across an organization becomes living context — continuously available to the people and AI that need it.
Five beliefs we started with — and hold harder now than when we wrote them down.
01
Management does not scale.
The cost of coordination grows exponentially while the organization grows linearly. It's not a process problem. It's a math problem.
02
Systems of record can't keep up.
Every record drifts from reality the moment it's written. None of it is the truth. All of it is a performance of the truth.
03
Meetings are where the magic happens.
Strategy is set in conversations, not documents. Yet 95% of what happens in those rooms never leaves them.
04
AI is only as useful as the context it can access.
A copilot or agent can't act on what it can't see — what changed, what was decided, what matters now. A better model doesn't fix missing context.
05
Shared context is the infrastructure.
A living, governed layer for your people, your tools, and every AI agent in the stack. The latest, not the last-recorded.
→
We no longer ask if companies want this.
We ask why any organization wouldn't. Read the argument in full — both volumes are above.
Workpages · You write Version Two
The rest of the book is not ours to write.
Vol. II ends with twelve questions and the right-hand pages left blank. Answer them here — we read every response, and Version Two will be assembled from what you send.
12 questions · all optional · named or anonymousConnect them to shared context and they stop being assistants and start being colleagues.
— The Coordination Tax, Vol. I