Stockertim Dictionary · Artificial intelligence
Human checkpoint
Also called approval step, human review step
In short
A human checkpoint is a fixed point in a workflow where a named person must review and approve the work before it moves to the next step or leaves the team.
- Beginner
- Concept
- Checked
In detail
A human checkpoint is a deliberate pause in a process. Work produced by a person, a tool or an AI assistant stops at that point until a named reviewer has looked at it and approved it. Only then does it move on, whether to a customer, a payment run, a published page or the next stage of a project.
A checkpoint is defined by four things: where it sits in the workflow, who the reviewer is, what they must check and what happens if they reject the work. Writing these down turns a vague intention to keep an eye on things into a real control that can be audited.
Checkpoints belong where an error would be costly or hard to undo: before anything is sent outside the organisation, before money moves, before records are changed and before decisions about people. In AI-assisted workflows they are the main way of keeping a person accountable for the output.
A checkpoint only works if the reviewer has the time, information and authority to say no. If approval becomes a click with no reading, automation bias takes over and the control exists only on paper. Some teams track how often reviewers return work as a simple test that the checkpoint is real.
Worked example
Northwind Bikes uses an AI assistant to draft replies to warranty claims and adds a checkpoint before any reply is sent.
- The workflow map marks the checkpoint after drafting and before sending, with the service manager as reviewer.
- The reviewer checks three things: the claim details match the order record, the reply quotes the real warranty terms, and any refund amount is correct.
- Rejected drafts go back with a note, and the manager records how many were returned each week.
No warranty reply leaves the business without a named person's approval, and the weekly return count shows the checks are real.
In practice
Human checkpoints appear in approval workflows, publishing systems, payment runs and AI-assisted customer service. They are often written into acceptable use policies and procedures with a named role rather than a general instruction to review.
Watch out
- A checkpoint without time or authority to reject is a rubber stamp, not a control.
- Human in the loop describes the overall design; a human checkpoint is one specific approval point within it.
Learn it properly
Official sources
- AI Risk Management Framework · National Institute of Standards and Technology
- NIST AI RMF Playbook · National Institute of Standards and Technology
Facts last checked on 3 November 2026.
Education, not advice: the Stockertim Dictionary explains what terms mean. It never recommends buying or selling anything.