AI & Technology
More Copilot, Slower Writing
Why does a sales team that added Microsoft Copilot to Outlook see draft turnaround times double, even as email accuracy improves?
2026-07-301 min read
The paradox stems from what I call the Automation Amplification Loop: a tool that speeds a sub‑task frees mental bandwidth, but that bandwidth is immediately re‑allocated to a new, higher‑order task that the tool does not yet handle. The result is a net slowdown in the original workflow. In the first quarter after Microsoft rolled out Copilot across its enterprise suite, a global consulting firm equipped 150 sales reps with the AI‑assisted draft feature.
Drafts that previously left a rep’s outbox in under a minute now lingered for three to five minutes as users paused to edit tone, verify data, and add strategic framing that Copilot could not generate. The extra cognitive step—“does this sound like me?”—became the bottleneck, even though the raw typing speed had increased.
The loop repeats: each efficiency gain invites a richer set of expectations, which in turn demands more human refinement, eroding the time saved. Over time, teams that recognize the loop restructure their processes, delegating the new refinement step to a separate role or to a second‑generation AI, thereby restoring the net gain.
Key insights
AI that automates a narrow step often creates a new, higher‑order step that humans must perform manually.
The net time saved only materializes when the new step is itself automated or off‑loaded.
Why it matters
Ignoring the loop lets hidden delays accumulate, turning an AI “speed‑up” into a productivity drain that hurts revenue cycles.
The loop also inflates cognitive load, increasing fatigue and error rates in tasks that were previously routine.
Use this tomorrow
1Open the latest Copilot‑generated email draft in Outlook, count the number of times you press “Edit” before sending, and note whether that count exceeds one.
2In your CRM, compare the average time‑to‑first‑response for contacts handled with Copilot versus those without, looking for a rise of more than a minute.
Go deeper
The term “Automation Amplification Loop” builds on classic “productivity paradox” research, notably Robert Solow’s observation that computing advances didn’t immediately raise output. Modern AI magnifies the effect because it reshapes the granularity of work: what once was a single manual act becomes a two‑part chain—AI generation followed by human curation.
The loop is especially pronounced in knowledge work where style, context, and persuasion matter. As AI improves at factual synthesis, the remaining human contribution shifts toward narrative authority, a skill that is slower to automate and often requires dedicated expertise.
