The deliverable is the load-bearing wall of corporate life — and AI is about to tear it down. The atomic unit of business shifts from deliverables to decisions.
The deliverable — memos, decks, analyses, reports, briefs, models, plans — is so deeply ingrained in business that it goes unnoticed until it breaks. It serves as proof of work, justification for headcount, and the basis for careers.
Human attention is limited and writing is slow. A CEO cannot consume raw data — someone compresses it into transportable thinking. The deliverable is "a compression algorithm for institutional knowledge." 60% of knowledge workers' time goes to communication and documentation (McKinsey).
Claude, Gemini, and GPT can each produce a competent first draft of a ten-page strategy memo in minutes for under a dollar in API costs. Quality is approaching "very good to great" — universally available. When production is free, counting it as value becomes absurd.
With 25+ years of strategy workshops, the author notes the best meetings involved "dialectic discussion and Socratic debate about what to do." Those were rare due to enormous preparation costs. AI drives those costs toward zero.
Memos, decks, reports — the unit of production that justifies headcount, drives performance reviews, and forms the backbone of organizational structure.
Claude, Gemini, GPT produce competent first drafts in minutes for under $1. Quality approaches "very good to great." The cost of deliverable production collapses toward zero.
Metrics shift from production cost to decision quality — encompassing data infrastructure, AI tools, human judgment, and process quality.
Reviews built around counting deliverables become absurd when production is commoditized. "The only review that matters is the quality of decisions made."
Management layers that exist to review and route deliverables must find new purpose when AI produces drafts that go directly to decision-makers with source attribution.
With decisions as the atomic unit, failure ownership shifts from shared production toward radical individual accountability — the final decision-maker is exposed.
Deliverables do more than inform decisions — they create shared context, build institutional memory, and force rigorous thinking through writing. If AI writes the memo, you get the artifact without the cognitive work.
"Writing is how an analyst discovers what they think. If AI writes the memo, you get the artifact without the cognitive work. The risk is optimizing for decision speed while losing the deliberative quality that slow, human-produced deliverables once enforced."
— Shelly Palmer, May 2026
Honestly identify which deliverables were always busywork — reports nobody reads, models confirming predetermined decisions, decks serving political purposes — versus those doing cognitive work AI cannot replicate.
Preserve the cognitive work while outsourcing mechanical production. Use AI to challenge assumptions and surface counterarguments. The goal: "preserve the cognitive work while outsourcing the mechanical production." Never let AI replace the thinking.
Proof of work, transportable thinking — the load-bearing wall of corporate life.
The new atomic unit — what remains after AI commoditizes deliverable production.
Smallest irreducible unit of work defining measurement, reward, and structure.
Legacy metric approaching zero as AI commoditizes production — becoming obsolete.
Replacement metric: infrastructure, AI tools, human judgment, process quality.
Unread reports, political decks — becomes visible when production is free.
Final decision-maker owns the outcome — no deliverable shield.
Shared context built through writing — the counterargument's core concern.
Catalog every recurring deliverable. Ask: does anyone read this? Does it inform a real decision? Or is it busywork?
Separate deliverables where writing IS the thinking from those where writing just documents conclusions. Outsource the latter to AI. Preserve the former.
Move budget from deliverable headcount toward decision-support infrastructure and decision-quality assessment.
Evaluate decision quality, not deliverable counts. Assess reasoning, counterargument handling, and outcome correctness.
Identify routing/review layers made redundant by AI. Transition talent into decision-quality or decision-support roles.
Use AI to challenge assumptions, surface counterarguments, produce drafts for human debate. Never let AI replace the thinking.
Reward decision quality and intellectual honesty. Recognize this transition from 'looking busy' will be terrifying for many.