Agent Delegation as Job Crafting: Letting Employees Choose What to Automate
Choosing what to delegate to AI can become a form of job crafting, allowing employees to reshape task boundaries around higher-value work—if automation remains voluntary, visible, and development-aware.
Delegation to agents can become a new way employees craft their jobs
Job-crafting theory describes employees as active shapers of the task, relational, and cognitive boundaries of their work.[1] Agentic AI creates a new lever for that process: workers can decide which recurring tasks to hand off, which responsibilities to keep, where human collaboration matters, and how to reinterpret their role as automation changes the workflow. That choice can increase motivation because the employee is not merely receiving a redesigned job from above; they are participating in the redesign. The important condition is real discretion. If a central automation team silently removes tasks and redistributes the rest, the organization may be automating work without enabling job crafting.
Automation can be participatory or imposed
The same technical change feels different when employees help choose it. Participation gives workers a chance to protect developmental tasks, meaningful relationships, and responsibilities that define professional identity.
Job crafting should optimize meaning and capability, not just convenience
Self-Determination Theory in work organizations emphasizes autonomy, competence, and relatedness.[2] Delegation can support autonomy when employees choose it, competence when it frees time for harder work, and relatedness when it reduces administrative burden around collaboration. It can undermine all three if workers delegate the parts of the job that helped them feel skilled, independent, or connected to others. A good “what should I automate?” conversation therefore includes more than time saved. It asks which tasks build expertise, which tasks create purpose, which tasks connect the employee to customers or teammates, and which tasks are genuinely low-value repetition.
Autonomy-supportive work design suggests employees need voice over methods and priorities
Meta-analytic research on autonomy-supportive leadership links such support with autonomous motivation, well-being, need satisfaction, and positive work behavior.[3] Agent delegation can operationalize that principle. Employees might be given an automation budget or menu of agent capabilities, then choose where to apply them within safety boundaries. They can also specify preferred review depth and decide which tasks should remain manual for learning. This is stronger than a one-time “opt in” because it recognizes that delegation preferences change as a worker becomes more experienced or as the job evolves.
Delegation should be revisable
A task that was worth practicing last quarter may be routine today. A task automated too early may need to be brought back for skill development. Workers should be able to change the boundary.
The agentic workplace makes delegation a core professional skill
Microsoft’s 2026 Work Trend Index describes a workplace in which agents take on execution while people direct work, make decisions, and own outcomes.[4] In that model, deciding what to delegate is itself part of expertise. Good delegation requires understanding task structure, risk, uncertainty, dependencies, and the quality of available automation. Organizations should teach employees how to decompose work into “automate, assist, review, keep human” categories instead of treating AI use as a binary choice. This makes the employee an architect of the workflow rather than a passive consumer of automation.
Evidence from generative AI at work suggests augmentation can help workers climb the learning curve
The NBER customer-support study found average productivity gains and particularly large benefits for novice and lower-skilled workers, alongside suggestive evidence of learning.[5] That supports a model in which delegation and assistance are development-sensitive. Early in a role, an employee may use AI to learn patterns while retaining more hands-on practice. Later, they may automate those same routines and redirect effort toward complex cases. Job crafting should therefore evolve with mastery. Uniform automation policies can miss this progression because they assume the same task has the same motivational and developmental value for everyone.
The right delegation boundary depends on career stage
A routine task for an expert may be the practice arena where a newcomer builds pattern recognition. Systems should let role level and learning goals influence what is automated.
Teams can use a four-part delegation review: keep, assist, automate, redesign
A practical job-crafting workshop can classify tasks into four groups. Keep tasks that are meaningful, developmental, or require human judgment. Assist tasks where AI can provide suggestions while the employee remains active. Automate tasks that are repetitive, reversible, and well understood. Redesign tasks whose current process exists only because of old constraints. This prevents automation from becoming a race to eliminate clicks. Sometimes the best answer is not to give the task to an agent but to remove the task entirely or change how the team coordinates around it.
Managers should protect equitable access to job crafting
If only senior or technically confident employees can redesign their work with agents, AI may widen opportunity gaps. Organizations should provide training, templates, safe experimentation environments, and time for workflow redesign. Managers can also review whether people are being left with disproportionately tedious or emotionally difficult tasks after automation. Delegation choices should improve the quality of work across the team, not simply allow a few employees to offload undesirable tasks onto colleagues or automated systems that create new hidden labor.
Job crafting is not a privilege for power users
The opportunity to shape one’s work should be distributed. Otherwise agentic tools can create a two-tier workplace in which some employees gain leverage while others inherit the leftovers.
The best delegation strategy leaves the employee with a stronger, more meaningful role
Agent delegation becomes motivating when it feels like an expansion of professional agency. The worker can choose what to offload, preserve practice where it matters, maintain relationships that give the job meaning, and take on responsibilities that better use human judgment. Organizations should make delegation transparent, revisable, equitable, and aligned with career development. The goal is not to keep humans busy for the sake of employment theater. It is to redesign the role so that the remaining human work is worth owning. When employees help draw the boundary between human and agent, automation can become a form of job crafting rather than a form of job erosion.
Works Cited
- 01Academy of Management Review — Crafting a Job journals.aom.org
- 02
- 03PMC — Leader Autonomy Support in the Workplace: Meta-Analytic Review pmc.ncbi.nlm.nih.gov
- 04
- 05NBER — Generative AI at Work nber.org
CodeHistory is a living archive. Citations document the evidence used for this edition; later evidence may refine the account.