Algorithmic management: AI that allocates and steers work, beyond platform work

Short answer: Algorithmic management is AI that allocates work, steers performance and influences behaviour — and it is long past being limited to delivery and taxi platforms. An ordinary company that uses software to assign shifts, score productivity or "nudge" staff is also doing algorithmic management. In the AI Act this falls under Annex III (task allocation and performance evaluation); the GDPR and — for platforms — the Platform Work Directive add to it.

What it is

Algorithmic management covers systems that: allocate tasks or shifts based on behaviour or traits; measure and score performance in real time; steer work pace or routes; and influence behaviour via notifications, scores or "gamification". The common thread: the system makes or steers decisions a manager would normally make.

Why it is high-risk

Annex III classifies both task allocation based on behaviour/traits and performance evaluation as high-risk. It does not matter whether it is couriers or office staff. So the deployer's duties apply (Art. 26): human oversight, information to workers and monitoring for skewed outcomes.

The Platform Work Directive as guidance

The Platform Work Directive regulates automated management strictly: transparency about the systems, a ban on certain data processing, and human oversight of important decisions. Formally it applies to platform work, but it sets the norm for what "decent algorithmic management" is — guidance ordinary employers can follow too.

The trap: steering without oversight

The risk sits in the sum: separate scores and notifications that together dictate the work, without anyone still assessing the outcome. Then the system effectively steers the employment relationship. Combine this with monitoring: measuring that turns into steering is exactly where high-risk begins.

What to do

• Inventory where software allocates, scores or steers tasks — not just your "HR tools".
• Build human oversight into decisions that affect workers.
• Be transparent about which systems steer the work and how.
• Involve the works council — steering systems engage the consent right.

Algorithmic management promises efficiency but shifts power to the system. The law requires a human to stay at the helm — even when the software does most of the steering.