The deliverable is the atomic unit of business. The memo, the deck, the analysis, the report, the brief, the model, the plan. These artifacts are so deeply embedded in how we work that we’ve stopped noticing them, the way you stop noticing the foundation of a building until it cracks. Deliverables are how knowledge workers prove they worked, how managers justify headcount, how executives learn about the day-to-day without reading the raw data, and how entire careers are built and evaluated. The deliverable is the load-bearing wall of corporate life and AI is about to tear it down.
Why the Deliverable Exists
The deliverable exists because human attention is limited and human writing is slow. A CEO cannot read every customer complaint, every market report, and every competitive filing. Instead, someone compresses that information into a five-page memo, a ten-slide deck, or a one-page brief. The deliverable is transportable thinking, a compression algorithm for institutional knowledge that converts hours of reading into minutes of consumption.
McKinsey found that knowledge workers spend roughly 60 percent of their time on communication and documentation, with only 40 percent devoted to the actual expertise they were hired for. We built entire organizational layers, career ladders, review cycles, and departmental headcount around the production of these compression artifacts. It was the only way humans could scale an organization.
Enter AI
Claude, Gemini, and GPT can each produce a competent first draft of a ten-page strategy memo in minutes for less than a dollar in API costs. Soon the quality range will be very good to great and, most importantly, the technology will be available to everyone. As agents (AI acting on your behalf) assume more responsibility for deliverables, the atomic unit of business will shift.
The Atomic Unit Shifts
I’ve been running strategy, roadmap, and data-driven-decisioning workshops with C-suite teams for over twenty-five years. The best meetings I’ve ever been in had one thing in common: The sessions were focused on dialectic discussion and Socratic debate about what to do. Those meetings were rare because the preparation cost was enormous. AI makes that preparation cost approach zero. The atomic unit of business must inevitably shift from “the deliverable” to “the decision.”
What Breaks
1. Performance Reviews
Reviews are built around deliverables: How many reports did you produce? How polished was your deck? When the deliverable is commoditized, evaluating people on production volume is like evaluating factory workers on how many widgets they assembled by hand after you installed robots on the line. The only review that matters is the quality of decisions made.
2. Organizational Structures
Org charts break because entire layers of management exist to review, approve, and route deliverables up the chain. When an AI agent can produce a draft that goes directly to the decision-maker with full source attribution, the review layer needs a new reason to exist.
3. The Accountability Gap
If the atomic unit is the decision, who “owns” a failure? In the old world, the author of the memo shared the blame. In the new world, as we shift from shared production to radical individual accountability, the executive making the final call stands alone.
Cost per Deliverable → Cost per Decision
If we start from first principles, we can financially measure every business function by Cost per Deliverable. When AI collapses the cost of production, the fundamental financial measurement will evolve into Cost per Decision. How much does it cost your organization to make a good strategic call? That cost includes the data infrastructure, the AI tools, the human judgment applied, and the quality of the decision process. Companies that measure themselves this way will allocate resources very differently than companies that still literally (or figuratively) count deliverables produced and delivered.
The Busywork Reveal
Almost everyone has written a report that went into a drawer, built a model that confirmed a decision already made, or prepared a deck for a meeting that was really about politics. When production becomes free, busywork becomes visible. If an AI can produce the report in thirty seconds and no one reads it, the fiction that the report mattered is harder to maintain.
We are moving from an era of “looking busy” to an era of “being right.” For many corporate cultures, this is a terrifying promotion.
How I Might Be Wrong
The strongest counterargument is that deliverables do more than inform decisions. They create shared context, build institutional memory, and force the kind of rigorous thinking that only comes from having to put something in writing. A strategy memo isn’t just an input to a decision; the act of writing it is how the analyst discovers what they actually think. This is a critical training function for entry-level knowledge workers.
If AI writes the memo, you get the artifact without the thinking. The deliverable might be load-bearing in ways that survive commodification: as a coordination mechanism, a record of reasoning, and a forcing function for clarity. The risk is that we optimize for decision speed and lose the deliberative quality that slow, human-produced deliverables once enforced.
Next Steps
I take this seriously. There are two next steps that will help prepare us for what’s coming. First, we have to figure out which deliverables have always been busywork and which ones were doing cognitive work that AI cannot replicate. Then, we need to start treating AI as a sparring partner, not as a ghostwriter. The goal is to preserve the cognitive work while outsourcing the mechanical production.
As best I can tell, most companies aren’t ready for this level of honesty. The economics of AI are as much about new ways to organize our thinking as they are about new ways to organize production.
Every company needs a Claw strategy. Do you have one?
Author’s note: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it. This work was created with the assistance of various generative AI models.