When work piles up, the old instinct is to hire. AI adds a second option - automate - and the interesting question is no longer "who do we hire?" but "should we hire or automate this?" Answered case by case, it is one of the more valuable decisions a leader now makes. Here is a framework that keeps it honest.
Start with the task, not the headcount
Break the growing workload into its actual tasks, and judge each on four questions:
- Is it repetitive and rules-based? The more so, the better it suits an agent.
- Does it need judgement, relationships or creativity? The more so, the more it needs a person.
- How variable are the inputs? High variety favours people; consistency favours automation.
- What does a mistake cost? Expensive, hard-to-reverse errors need human oversight.
Most real workloads are a mix - which usually points to both: automate the routine portion, hire for the judgement portion.
Run the honest numbers
Automation looks cheaper until you count setup, oversight and maintenance; hiring looks dear until you count the flexibility and judgement a person brings. Compare the true cost of each for the specific task, using the framing in AI agents vs hiring more staff - not headline prices.
The framework in one line
- Automate repetitive, consistent, low-stakes, high-volume work
- Hire for judgement, relationships, variety and high-stakes decisions
- Usually, do both - an agent takes the routine so a smaller, higher-value team handles the rest
Avoid the two traps
Do not automate work that really needs a person (you damage quality), and do not hire for work an agent could do better and cheaper (you waste money). The skill is telling them apart - the same lens as scaling without hiring proportionally.
The bottom line
Decide task by task: automate the repetitive and consistent, hire for judgement and relationships, and usually combine the two. Learning to make that call well is part of the AI Strategy for Executives course at London School of Business UK. Enquire today.