Identify the acting node
Disclose whether a recommendation or decision came from a human, assisted human, bot, AI assistant, agent, organization, or mixed actor.
Bounded application profile
VNWO for AI-mediated work systems
Static templates for making algorithmic assistance, delegated authority, consequential work, memory use, review, repair, and exit inspectable without turning VNWO into workplace monitoring, labor-law advice, or a certification program.
Purpose
Make algorithmic work decisions reviewable without normalizing worker surveillance.
AI-mediated work systems can combine scheduling, ranking, communications, memory, endpoint access, quality review, and offboarding. VNWO applies its existing observatory patterns to these systems so people can see who acted, what data was used, what authority applied, what consequence followed, and how review or exit works.
The profile is useful to platform operators, worker representatives, vendors, governance reviewers, and procurement teams as a documentation baseline. It is not legal advice, employment-policy certification, live monitoring, workforce telemetry, or a substitute for collective bargaining and jurisdiction-specific review.
Controls
Seven public surfaces for AI-mediated work
Bound delegated authority
State which scheduling, ranking, review, communication, or access actions the system may propose or perform, and which require a human decision.
Record consequential work
Create a receipt when automation materially affects scheduling, access, visibility, compensation inputs, discipline workflow, or offboarding.
Disclose memory and retention
Name the records used, their retention period, correction path, secondary-use limits, and retirement or export route.
Provide human review and repair
Publish how a person can challenge incorrect data, an automated recommendation, a refusal, or a policy application.
Use no-op under uncertainty
Defer or refuse an automated action when evidence, authority, review capacity, or source provenance is insufficient.
Preserve disconnection and exit
Document off-hours boundaries, delegation revocation, record export, account closure, and retention after separation.
Fictional example
Shift recommendation with a human decision boundary
This example illustrates the documentation pattern only. It does not represent a deployed employer system or verified legal process.
| Node | Scheduling recommendation assistant operated by Example Cooperative. |
|---|---|
| Permitted work | Propose a shift assignment from published availability and qualification records. |
| Prohibited work | Final discipline, compensation change, or termination without an identified human decision maker. |
| Evidence | Availability window, required qualification, reason code, source version, and conflicting-data note. |
| No-op trigger | Conflicting availability, missing qualification evidence, expired consent, or insufficient human review capacity. |
| Review | Named human reviewer can correct records and issue a replacement decision receipt. |
| Exit | Revoke automated delegation, export relevant records, document retention, and provide a correction path after separation. |
Adoption
Publish the evidence surface before relying on the automation.
- Name the acting system and responsible operator.
- Publish allowed and prohibited authority.
- Map data and memory sources with retention and correction paths.
- Define material decisions that require a receipt.
- Define no-op and human-review triggers.
- Test appeal, revocation, export, and exit before launch.
Required source routing
- Route legal and employment obligations to qualified jurisdiction-specific review.
- Route memory-package standards to UAIX.
- Route theory claims to Teleodynamic.
- Route endpoint capability and authorization boundaries to LocalEndpoint and the implementing system.
- Route incidents and recovery evidence to the implementing incident process or ErrorNotifier lane.