Classification & AI Boundaries
See what ZoikoTime calculates, what approved AI may assist with, what Kairos can explain, and where an authorized person must decide.
.png)
Standard - 8h Active
Rule matched: USA_CA_Overtime_V2
Systems can calculate, flag, retrieve and explain. Authorized people review and decide consequential outcomes.
The direct answer
ZoikoTime keeps time classification deterministic and reviewable. Approved machine-learning functions, where currently authorized, may support anomaly or signal-quality review; they do not make payroll, discipline, misconduct, legal, or other consequential decisions. Kairos retrieves, summarizes, and explains governed data and decides nothing. When a record needs judgment, correction, approval, or escalation, an authorized human remains responsible and the decision history is preserved.
Three-Lane System Boundaries
Deterministic Classification
Apply approved policy-bound rules, expose rule trace, show limitations.
AI/ML inference, probabilistic judgment, hidden model behavior.
Rule trace, policy version, inputs, output.
Approved AI Assistance
Surface anomaly/signal-quality info to support review.
Consequential decision, worker ranking, punitive score.
Signal type, purpose, version/status, limitations.
Kairos
Retrieve, summarize, explain authorized governed data.
Classify, approve, decide payroll/discipline/legal.
Sources, freshness, scope, limitations.
Review, correct, approve, reject, escalate and decide within role authority. Consequential authority remains human and role-bound.
The Human Authority Lifecycle
1. Capture
2. Deterministic Classify
3. Optional Assist
4. Neutral Review State
5. Human Review
6. Authorized Decision
7. Notify / Reopen
8. Downstream Handoff
Visibility, correction, and human escalation
Understand what happened
Clear step-by-step breakdown of how your active hours were translated and categorized under active payroll rules.
Know whether AI was involved
Explicit signaling whenever machine-learning filters assist a manager in verifying context or flagging errors.
Correct facts/context
A transparent mechanism to submit modifications, adjust dates, or log actual physical hours worked.
Add context safely
Provide written notes or activity descriptions directly into the decision trace to clarify ambiguous logs.
Escalate to a person
Request formal human arbitration and override evaluation if system calculations feel mismatched.
See decision history
Trace who approved, modified, or verified your records across all historical reporting cycles.
See the proof
Review Queue
Classification Trace
Human Decision Record
Governance Directory & Core Documentation
Give the right people the right access — and prove it
See Classification & AI Boundaries configured for your identity providers, roles, approval workflows, and audit requirements.
30-day trial for eligible plans · Human authority remains controlling · Supported controls & assurances are documented in the Trust Center.