Audit-Ready AI for
Every Regulated Industry.
UnitMesh organizes AI governance around discrete, auditable decision units — each with built-in accountability (RACI), evidence-gated change management, and a complete chain of custody from input to recommendation.
Book a DemoDecision Units, Not Dashboards.
AI produces insights. What organizations struggle with is the last mile — translating insights into governed, auditable decisions with clear ownership and managed change.
UnitMesh is organized around David Pidsley's AI-Ready Decision Unit framework — the principle that human-AI collaboration works best when structured around discrete, auditable decisions with clear ownership and managed change. Each unit carries accountability from the moment it is defined, not as an afterthought.
Built on the philosophical and architectural concept of AI-Ready Decision Units by David Pidsley, Gartner's first Decision Intelligence Platform Leader. See also: Why Decision Intelligence Matters →
What the AI recommends
Each unit is scoped to a single, discrete decision — not a dashboard, not a score, not a report. The AI recommendation is bounded, testable, and auditable in isolation.
RACI accountability, built in
Every decision unit ships with a RACI map — who is Responsible, Accountable, Consulted, and Informed. Accountability is structural, not procedural. It cannot be disabled.
Chain of custody on every output
Every recommendation is wrapped in an immutable evidence record: model version, input snapshot, timestamp, and reviewer response. Audit-ready by construction, not by export.
Built for Your Industry
The same evidence infrastructure. The compliance language your regulators already speak.
- →Clinical AI decisions cannot be audited after the fact — no evidence trail, no regulator defense.
- →Model performance degrades silently — no systematic monitoring between audits.
- →EHR integration requires point-in-time data that can be reconstructed on demand.
UnitMesh activates clinical decision units through a governed shadow-to-active pathway. Every recommendation carries a full evidence chain: model version, feature snapshot, timestamp, and clinician response — all linked by deterministic evidence identifiers. When a regulator asks "why did the AI say that?" you have a complete answer.
AI governance where every clinical decision traces back to verifiable evidence.
The Four Clinical Decision Units
Each unit runs in shadow mode, passes formal audit, and requires dual or triple sign-off before going live.
CHF Discharge Routing
Evidence-based discharge tier recommendation for congestive heart failure patients — reducing preventable readmissions.
Medication Reconciliation
Automated reconciliation routing at transitions of care, flagging high-risk discrepancies for pharmacist review.
ICU Readmission Risk
Post-ICU readmission probability scoring, enabling proactive step-down and early intervention.
Sepsis Early Warning
Continuous sepsis risk scoring with THRIL-composite alert routing to nursing and rapid response teams.
How Activation Works
Every decision unit follows the same governed pathway before a single recommendation reaches a reviewer.
Shadow Mode
90-day observation under live conditions. The AI generates recommendations in parallel with reviewers — no operational impact. RACI ownership is established before a single live recommendation is issued.
Formal Audit
Shadow period evidence is reviewed against threshold criteria. Dual sign-off required (Red-tier units add Legal). No unit advances without a closed evidence chain — the gate is structural, not procedural.
Active Deployment
Recommendations surface via FHIR CDS Hooks. Rollout follows the PROSCI ADKAR model — Awareness through Reinforcement. Every decision is logged with full provenance: model version, features, timestamp, reviewer response.
Healthcare Regulatory Pathway
UnitMesh is simulation-validated and tracking a defined regulatory pathway. Insurance/Climate and Pharma are in evaluation-stage partnerships.
3IT Health Network
3IT Health is the simulation reference network used to validate UNIT MESH — four decision units running across four hospital scenarios. The safety validation run completed 50K decisions with 0 adverse events and demonstrated federated model improvement across activation stages.
The trust layer powering every decision — ProvenLake — is the same infrastructure available to any regulated-industry application built on 3it.dev.
Pricing
Start free and grow into full regulatory-grade evidence chains as your deployment matures.
- ✓Single industry vertical
- ✓Up to 500 verified decisions
- ✓30-day evidence retention
- ✓Internal team verification only
- ✓2 governed pipelines
- ✓Community support
- ✓Full chain-of-custody logging
- ✓Single industry vertical
- ✓Up to 10,000 verified decisions / month
- ✓12-month evidence retention
- ✓Internal audit access
- ✓Up to 5 governed pipelines
- ✓Email support (2-business-day SLA)
- ✓Decision Registry access
- ✓Performance monitoring dashboard
- ✓Up to 3 industry verticals
- ✓Unlimited verified decisions
- ✓36-month evidence retention
- ✓External auditor access portal
- ✓Unlimited governed pipelines
- ✓Priority support (next-business-day SLA)
- ✓Formal audit workflow support
- ✓Multi-approver sign-off chains
- ✓Custom evidence retention policy
- ✓All industry verticals
- ✓Unlimited verified decisions
- ✓Custom retention (up to 10 years)
- ✓Regulator-facing evidence export
- ✓Unlimited governed pipelines
- ✓Dedicated support with named CSM
- ✓On-premise or private cloud deployment
- ✓Custom approval chain configuration
- ✓Federated multi-region deployment
- ✓SLA negotiated per contract
Book a Demo
See UnitMesh in action. We'll reach out within one business day.