Leadership1 publisher3 min readPublished
Before hiring, sort the role's ten tasks into four buckets to decide whether to hire, automate or redesign
An Entrepreneur contributor column asks managers to list the ten recurring tasks behind a requisition and sort them into judgment, relationship, repetition and coordination work. Its supporting studies measured support agents and consultants.
The Board Room · Leadership desk
What happened
- An Entrepreneur contributor column tells managers to write down the ten recurring tasks they expect a proposed hire to do, before the title or the qualifications are discussed.
- Its rule sends judgment and relationship bottlenecks toward hiring or developing a person, and repetition bottlenecks toward an automation test before any permanent headcount is added.
- Coordination bottlenecks get a process redesign first, and a role containing all three is split so the human part of the job is built around the work that deserves a human.
- The U.S. Census Bureau put overall business AI use at about 17% to 20% across late 2025 and early 2026, with larger firms and knowledge-intensive sectors adopting at much higher rates.
- An NBER study of more than 5,000 customer support agents found a 14% average productivity increase from AI assistance, with substantially larger gains for less experienced workers.
Compiled by The Board RoomSomething wrong?How this is made
Why it matters
- decision The sequence changes which decision comes first. Approving the person before the redesign gives the disconnected handoff an owner, and the automation test that would have priced the alternative never runs.
- constraint The productivity evidence limits what an automation case can promise: it was measured on support agents and consultants inside AI's capabilities, so it does not underwrite claims about judgment-heavy roles.
- contradiction A manager's sense of being behind depends on which base was counted. The column cites both the Census range and the Stanford 88% without reconciling the populations behind them.
Coordination is the most informative of the four buckets, because of how the column defines it: work that exists because systems or people are disconnected, including copying information between tools, chasing approvals, reconciling versions and asking the same status question repeatedly [5]. That work is nobody's job by design. It accumulates in the gaps between systems, and it reaches a requisition because the nearest team is the one absorbing it. For that bucket the instruction is to redesign the process before automating anything [4].
Automation amplifies process design, the column argues, so where ownership is fuzzy and the data is wrong, automation makes the confusion faster [11]. The same logic applies to a person. Put someone in the seat and the broken handoff acquires an owner and a calendar.
The 14% average from the NBER support-agent study is the nearest thing in the record to a capacity figure. Held at that average, a team of seven picks up capacity equal to one more worker [15]. Two caveats sit on it. Those were customer support agents, and the Harvard Business School field experiment found consultants gaining in speed and quality on tasks inside AI's capabilities while performance suffered outside that boundary [10]. The figure describes assistance on repetition-heavy work, and the column's own definition of judgment work is that AI can assist but a named human should own the outcome [20].
The two adoption figures the column cites sit about 68 percentage points apart [16], and it does not reconcile the bases they were drawn from. Census diffusion research puts Sales and Marketing, Strategy and Business Development, and IT among the most common functions where adopting firms use AI [7]. Those are the same functions the column names as the source of headcount pressure in growing companies, through lead routing, follow-up, scheduling, reporting, data movement, content operations and customer communication [14]. The column's reading of the gap is that adoption is moving faster than operating-model redesign [17].
Much of this renames what a competent manager already does when writing a job description. The author half-concedes it, calling the rule a founder-level heuristic and writing that "It is not an industry standard, but it forces the right conversation" [13]. The piece runs under Entrepreneur's contributor label, which states that opinions are the contributors' own [19]. What the exercise changes is the timing of the analysis: the sort happens before the requisition is approved, while redesign and an automation test are still available.
The approval signed this quarter fixes the options available next quarter. Once the repetition and coordination tasks have a person attached, they are somebody's responsibilities and the process underneath them is funded at a recurring cost. The column puts it this way: "A job description should not be a storage unit for broken workflows." [12]
What to watch
- Whether the Census series moves off the 17% to 20% band as it updates through 2026. The band sets how far a small firm sits from the Stanford reading.
- A productivity study covering judgment-heavy roles would test whether the 14% support-agent result travels beyond repetition work.
- Stanford's next read on agent deployment, which the 2026 index still describes as early.