Model Cards
Our AI rules & guardrails require explainability. Here's every model we use, what it does, what data trained it, and how we govern it. Where a feature is switched off, or where no independent audit has been run, we say so and publish no score.
Governance Score
Composite score 0–100 across Explainability · Bias checks · Human review · Transparency.
Who's in charge
Each card says it plainly: suggestions only · a person approves · a person reviews exceptions · auto-blocks violations.
Guardrails
AI rules & guardrails enforced in this model's runtime path.
Syncora-native · v2.3
Suggests shift assignments balancing coverage, fairness, and predictive scheduling laws.
Training data
Anonymized scheduling data + state regulation features.
Override path
Manager approves every schedule.
Governance Score
AI Rules & Guardrails
Syncora-native · v1.7
Detects buddy-punching, time padding, geofence evasion, and overtime gaming.
Training data
Synthetic + opt-in customer data.
Override path
A person reviews every flagged event.
Governance Score
AI Rules & Guardrails
Syncora-native · v1.4
Predicts labor cost vs. revenue/demand 14 days out.
Training data
Aggregated labor + demand data.
Override path
Manager confirms before any auto-adjustment.
Governance Score
AI Rules & Guardrails
Syncora-native · v3.0
Continuous four-fifths-rule monitoring on lead scoring, scheduling, performance, and pay equity.
Training data
Rule-based + statistical. No ML.
Override path
Blocks decisions on four-fifths-rule failure.
Governance Score
AI Rules & Guardrails
Google · Gemini Live (gemini-2.5-flash native audio)
Spoken questions about Syncora — what it does, how it is governed, and how it is priced.
No score publishedSwitched off. We are moving the voice connection behind a secure server-side channel first. Nobody can use this feature today, so we publish no score for it. The text guide still works.
Tavus · conversational video persona, configured per deployment
Answers questions about using Syncora and can move you to the right page. It tells you it is an AI before you start.
No score publishedLive for signed-in administrators. No independent audit of this surface has been run yet, so we publish no score for it. It cannot read your employee data.
Syncora-native · v0.9 (beta)
Scores applicants for role fit; bias-firewalled per Title VII, NYC AEDT, and EU AI Act.
Training data
Anonymized hiring outcomes.
Override path
Scores never auto-reject; human always decides.
Governance Score
AI Rules & Guardrails
Syncora-native · v2.1
Detects unjustified pay gaps within and across protected classes.
Training data
Statistical. No ML.
Override path
Surfaces gaps for HR review; never auto-adjusts pay.
Governance Score
AI Rules & Guardrails
Syncora-native · v1.2
Generates personalized onboarding workflows from role + department.
Training data
Rule-based. No language model.
Override path
HR reviews each generated path.
Governance Score
AI Rules & Guardrails
Syncora-native · v1.0
Validates schedules against NYC, Oregon, Seattle, Chicago, and Philadelphia predictive scheduling laws.
Training data
Rule-based. No ML.
Override path
Blocks non-compliant schedules.
Governance Score
AI Rules & Guardrails
Syncora-native · v1.0
Enforces written-consent capture before biometric collection; auto-destroys biometric data per BIPA 740 ILCS 14/15(a).
Training data
Rule-based. No ML.
Override path
Collects only after consent; never bypassable.
Governance Score
AI Rules & Guardrails
Syncora-native · v1.0
Auto-drafts adverse action notices for AI-influenced decisions per Title VII, EU AI Act, and NYC AEDT.
Training data
Rule-based. Notice text is built from fixed wording plus the decision’s own figures. No language model.
Override path
HR reviews every notice before send.
Governance Score
AI Rules & Guardrails
Every decision these models make is logged — with its reasoning — where you can audit it. See our Trust Center, our public bias audits, and our complete sub-processor list. Questions? bob@cognitivewx.info
We use essential cookies to run this site and optional analytics cookies to improve it. You can accept all, reject non-essential, or read our Privacy Policy.