Senticit Slant™
AI ethics, bias & drift monitoring — built for healthcare.
A clinical AI tool that drifts from 94% to 81% accuracy is a seven-figure liability — in misreads, malpractice exposure, and accreditation findings — that nobody on your team is currently watching for. Slant gives you a defensible Slant Score™ across every model, every modality, every clinician, with accreditation-ready reports and HIPAA-grade deployment from day one.
Slant is the healthcare expression of our AI Integrity methodology.
The bias auditing, drift monitoring, ethics frameworks, and data provenance controls in Slant are the same disciplines we deliver to financial services, SaaS, and health-system clients through our broader AI Integrity & Resilience practice — productized, vendor-agnostic, and shaped to fit an accredited clinical environment.
See the underlying methodologyThree failure modes no one is watching
FDA, ACR, and Joint Commission are converging on AI governance as a formal accreditation requirement. Most practices have zero visibility into how their AI tools are actually performing.
Demographic Bias
AI tools trained on non-representative datasets quietly underperform for older, female, higher-BMI, or underrepresented patient populations. Invisible without active monitoring.
Model Drift
A tool that read at 94% accuracy at deployment may be at 81% eighteen months later as populations, equipment, and protocols evolve — and no one knows.
Ethical Deviation
Clinicians gradually defer to AI without critical review. Over-reliance is undetectable without systematic tracking of override rates and guideline adherence.
One number your board can defend
A composite 0–100 metric, updated in real time as your AI tools process cases. Translates statistical monitoring into a single signal clinicians, accreditors, and the leadership team all understand.
Bias Sub-Score
Demographic parity across age, sex, BMI, and ethnicity cohorts. Computed locally — PHI never leaves your environment.
Drift Sub-Score
Statistical comparison of current AI outputs vs. deployment baseline. Catches silent degradation before patient impact.
Ethics Sub-Score
Clinician override rates, guideline adherence, and consent documentation completeness from structured event logs.
Local-first AI engine.
Cloud LLMs on demand.
An open-source Ollama + Gemma 3 engine runs the majority of the workload on-premises or in your private cloud — keeping PHI inside your perimeter. Optional webhooks reach out to Claude, GPT-4o, Gemini, Azure OpenAI, or Perplexity only for narrative generation, cross-validation, or live regulatory lookups.
- PHI processed by local LLM — never sent to third parties
- Cloud webhooks send anonymized signals only
- Per-tenant Ollama instance available on Enterprise
- Bring-your-own-LLM endpoint supported
Data Flow
- 1IngestAI tool outputs arrive via API, FHIR, DICOM, or upload
- 2Local ScoreOllama + Gemma 3 computes bias, drift, ethics sub-scores
- 3Slant Score™Composite 0–100 written to your isolated Supabase tenant
- 4Optional EnrichCloud LLM webhook generates plain-language narrative
- 5Alert + ReportEmail, dashboard, accreditation pack
Pick the silo that fits your practice
Same engine. Same Slant Score™. Right-sized depth, ingestion, and governance for where you are today — with a clean path to where you're going.
Practice
Plug-and-play AI integrity
1–3 clinicians · solo or small group
Simple, guided setup with no IT lift. Get a defensible Slant Score™ on your AI tool within a day.
- Single AI tool connector + log upload
- Bias scanner & drift detector (local engine)
- Slant Score™ — composite 0–100
- Monthly Slant Report (PDF)
- 90-day score history & equity scorecard
- Email alerts when score drops
- Auto-generated HIPAA BAA
Group
Depth, collaboration, workflow
4–15 clinicians · multi-provider
Everything in Practice, plus ethics monitoring, multi-provider dashboards, and accreditation export packs.
- Up to 10 AI tools + FHIR (read-only) + DICOM metadata
- Ethics monitor, override tracker, confidence decay
- Cross-model consensus & per-modality splits
- Weekly Slant Report + accreditation pack (ACR / IAC / TJC)
- Provider comparison dashboards & trend reports
- Cloud LLM webhooks (Claude · GPT-4o · Perplexity)
- Up to 20 user seats · 48-hour SLA
Enterprise
Custom · scalable · governed
16+ · imaging centers · health systems
Multi-site aggregation, on-premise deployment, full EHR/PACS/RIS integration, and a named Senticit advisor.
- Unlimited AI tool connectors + Epic / Cerner / Meditech
- PACS / RIS / custom ETL pipelines
- Multi-site Slant aggregation & predictive drift
- Policy enforcement, consent tracking, SOC 2 evidence
- On-prem Docker, dedicated Ollama, custom data residency
- All 6 LLM webhooks + bring-your-own endpoint
- SSO/SAML · 99.9% uptime · 4-hour response · named advisor
What Slant watches for you
Bias Scanner
Demographic variance checks on every AI output. Local Ollama / Gemma 3 — no PHI leaves your environment.
Drift Detector
Statistical comparison of current outputs vs. deployment baseline. Catches silent degradation before it hits patients.
Slant Score™
Composite 0–100 integrity score, updated per batch. The single number your accreditors and board can trust.
Ethics Monitor
Detects over-reliance, guideline deviations, and consent gaps. Plain-language narratives via Claude webhook.
Cross-Model Consensus
Compares outputs across multiple AI tools and flags disagreement — your second opinion, automated.
Modality Splitter
Separate bias and drift tracking per modality — CT, MRI, X-ray, US — because one score doesn't fit all.
Override Tracker
Logs every clinician AI override with trend analysis. Surface the cases where humans are quietly correcting the model.
Custom Baselines & Cohorts
Per-site, per-department fingerprinting and org-specific demographic cohorts you define.
Regulatory Drift Predictor
ML-based forecast of when a tool is likely to breach a threshold — and a live FDA 510(k) certification tracker.
Reporting
- Monthly Slant Report (Practice)
- Weekly per-provider breakdown (Group)
- Accreditation pack — ACR · IAC · Joint Commission
- 12-month rolling trend analysis
- Executive & board-ready dashboards (Enterprise)
- REST API report export
Data Ingestion
- CSV / JSON log uploader
- Single API connector (Practice)
- Up to 10 AI tools + FHIR R4 + DICOM (Group)
- Nightly batch processor (Celery + Redis)
- Epic · Cerner · Meditech HL7 FHIR (Enterprise)
- PACS · RIS · custom ETL pipelines
Governance & Compliance
- Auto-generated HIPAA BAA (all tiers)
- Tamper-evident audit log viewer
- Policy enforcement engine (Enterprise)
- Patient consent tracking module
- AI incident lifecycle management
- SOC 2 Type II evidence package
Multi-model intelligence, your routing rules
| Webhook | Trigger | Purpose |
|---|---|---|
| Claude API | Ethics violation detected | Plain-language narrative generation |
| OpenAI GPT-4o | Bias score below 70 | Secondary validation cross-check |
| Perplexity | Regulatory alert | Real-time FDA / ACR guideline lookup |
| Gemini (Google) | Multimodal / DICOM metadata | Image-adjacent reasoning |
| Azure OpenAI | Enterprise HIPAA BAA workloads | Microsoft-compliant deployments |
| Custom webhook | Client-defined trigger | Bring your own LLM endpoint |
Side-by-side tier comparison
| Capability | Practice | Group | Enterprise |
|---|---|---|---|
| Bias detection | Basic | Multi-modality | Custom dimensions |
| Drift detection | Basic | Confidence decay | Predictive |
| Ethics monitoring | — | Full | Policy enforcement |
| Cloud LLM webhooks | — | 3 providers | All 6 + custom |
| Reporting | Monthly PDF | Weekly + accreditation pack | Custom builder + API |
| Data ingestion | File upload + 1 API | 10 APIs + FHIR basic | Unlimited + EHR / PACS / RIS |
| Multi-site | — | — | Full aggregation |
| On-premise | — | — | Full Docker package |
| User seats | 1 | Up to 20 | Unlimited |
| Support | Standard | 48hr SLA | Named advisor + custom SLA |
Deploy it your way
On-prem, private cloud, isolated LLMs, and enterprise identity — Slant runs where your compliance team needs it to run.
On-Premise Docker Package
Full Slant stack via Docker Compose, behind your firewall.
Private Cloud Deployment
AWS · Azure · GCP — your account, your VPC.
Dedicated Ollama Instance
Isolated local LLM, never shared with other tenants.
Custom Data Residency
Choose your region, your storage, your retention.
SSO / SAML Integration
Okta · Azure AD · Google Workspace.
Dedicated Senticit Advisor
Named technology advisory engagement/CISO point of contact.
Slant alone — or wrapped into a CTO + CISO retainer
Slant tiers above are the platform. Most healthcare groups bundle the platform with a Senticit professional services engagement retainer so an MD-aware CTO + CISO is accountable for the score, the accreditation packs, and the remediation plan.
Sentinel + Slant Practice
Solo / small practice (1–3 clinicians)
- Slant Practice platform included
- Quarterly Slant Score™ review
- Annual integrity audit & policy refresh
- Email-priority advisory
Vanguard + Slant Group
Multi-provider group (4–15 clinicians)
- Slant Group platform included
- Monthly executive Slant report
- ACR / IAC / TJC accreditation packs
- Bi-weekly leadership cadence
Summit + Slant Enterprise
Imaging centers, health systems (16+)
- Slant Enterprise platform included
- Multi-site Slant aggregation
- On-prem / dedicated Ollama deployment
- Named MD-aware CTO + CISO advisor
Bundles combine Slant platform pricing above with a Senticit technology advisory engagement + CISO retainer (Sentinel / Vanguard / Summit). Final pricing is shaped to your AI footprint during the discovery call.
Common questions
Talk to us about NIST & SOC 2 alignment
Get a direct conversation with our clinical AI compliance team. We'll show you exactly how Slant's monitoring, bias audits, and Slant Score™ map to NIST AI RMF functions and SOC 2 Common Criteria — and what your auditors will be handed.
Send us your clinical AI compliance question
Prefer not to book a meeting? Drop your NIST AI RMF or SOC 2 question below and our clinical AI compliance team will reply within one business day.
Ask a NIST / SOC 2 question
Get a written answer from our compliance team — usually within one business day. No call required.
Get the NIST AI RMF & SOC 2 Alignment Checklist
Built for clinical AI & healthcare leaders. A 6-section worksheet mapping your AI program to NIST AI RMF (Govern, Map, Measure, Manage) and SOC 2 Common Criteria — with the evidence artifacts auditors will request.
NIST AI RMF & SOC 2 Alignment Checklist
6-page PDF checklist with control mappings, evidence prompts, and an owner sign-off worksheet.
Pair Slant™ with Senticit AccredIQ™
The complete healthcare compliance + AI governance platform. Bundle Slant with AccredIQ — our accreditation readiness product — and save 20% on both subscriptions for the life of the contract.
Find out your Slant Score™
Book a 30-minute walkthrough. We'll connect to one of your AI tools live and show you your first Slant Score™ before the call ends.
Book a Slant DemoAccess tiers
What's included at each tier
Start free. Upgrade when you need depth, history, or named-CISO sign-off.
| Senticit Slant™ | Public Free | Prospect Free | Sentinel Contact us | Vanguard Contact us | Command Contact us |
|---|---|---|---|---|---|
| Product overview & demo request | |||||
| Bias detection across subgroups | |||||
| Continuous drift monitoring | |||||
| Accreditation-ready reports (ISO 42001, NIST AI RMF, EU AI Act) | |||||
| Radar↔Slant live sync | |||||
| White-label deployment |