How Our AI Technology Works
The intelligence behind Intelligent Roster®—built for the complexity, nuance, and human reality of healthcare rostering.
AI Built for Healthcare,
Not Generic Scheduling
Intelligent Roster® uses a dual-AI setup designed for healthcare: one model for deep, multi-step reasoning and one for fast, high-volume tasks—so you get both thoughtful decisions and quick answers.
Deep Reasoning
Multi-constraint roster generation, fairness across multiple dimensions, coverage impact, rule and conflict checks, natural-language rule handling, and staff–shift fit. This is the “why” and “how” behind each suggestion.
Fast Operations
Quick lookups, coverage summaries, shift breakdowns, automated reports, and routine data questions. This is the “doing”: fast, reliable, high-frequency use—without slowing down day-to-day questions.
The Dual-AI Advantage
We don’t use a single “one size fits all” model. We use two, each where it’s strongest. Roster building needs both careful reasoning and fast answers.
Claude
Deep Clinical ReasoningHandles the thinking. Roster generation, fairness analysis, coverage impact, hard-rule and conflict checks, understanding your department’s rules and structure, staff–shift compatibility.
- Roster generation & optimisation
- Fairness analysis across 7 dimensions
- Coverage impact modelling
- Hard-rule & conflict checking
- Staff–shift compatibility
- Natural language rule interpretation
GPT
Speed & VolumeHandles the doing. Quick searches, coverage summaries, shift breakdowns, report generation, routine data interpretation—fast and reliable for high-frequency use.
- Quick roster lookups & queries
- Coverage summaries
- Shift breakdowns
- Report generation
- Routine data interpretation
- Fast, reliable, high-frequency responses
Infrastructure You Can Rely On
AI Credits
Generous monthly allocation, clear usage tracking, automatic refunds on failed calls, 12-month rollover.
Rate Limiting
Per-organisation limits so no single tenant can affect system stability or drive runaway cost.
Security & Privacy
Encrypted keys, multi-tenant isolation, healthcare-privacy minded design, no data shared between organisations.
Monitoring
Real-time analytics, performance visibility, and automatic failover where applicable.
Ask IRIS Anything, Naturally
IRIS understands your streams, craft groups, rules, and language. Ask in plain English — no commands or menus needed. It’s like having a rostering expert available 24/7.
Roster Builders & Admins
Staff (Their Own Data Only)
From Requirements to Roster
in 5 Steps
Unlike systems that only “fill slots,” Intelligent Roster® is built around a five-level hierarchy. The AI understands clinical, operational, and human constraints — not just empty cells.
Understand What’s Needed
The system maps what your department needs: shift patterns, vacancies and coverage requirements, craft groups and scopes, streams, date ranges, and service constraints. The AI knows what shifts exist and how many people are needed, in which roles.
Levels 1 & 2Build the Availability Pool
The AI combines staff’s own availability (highest priority), master roster patterns, approved and pending leave, and qualifications and scope restrictions — giving a live picture of who can work when, not just who is “on the list.”
Levels 3 & 4Apply Your Organisational Rules
Rules are applied in a clear order: organisation-level rules, then stream, then craft group, then individual preferences. Every rule is respected; every conflict is flagged. The AI doesn’t bypass your policies — it works within them.
Rule EngineOptimise Who Goes Where
Using constraint-aware logic, the system balances availability, fatigue and rest rules, consecutive days, weekend load, holiday distribution, total hours, scope matching, and fairness across multiple dimensions. The roster isn’t just filled — it’s balanced and explainable.
Fairness EngineDeliver the Roster
In under 30 seconds, you get: a complete, optimised roster; any conflicts with reasons; coverage gaps; fairness impact; and AI reasoning so you can see why suggestions were made. Transparent, explainable, and auditable.
Level 5 — Living RosterWhat Makes Our AI Different
Clinical context, healthcare constraints, and genuine fairness — built in from the ground up, not bolted on.
Craft Groups & Scopes
Correct roles (e.g. RN, EN, doctor) and qualifications (e.g. resuscitation, trauma) matched to each shift automatically.
Seniority & Skill Mix
Respects your departmental structure and skill-mix requirements — so every shift has the right people.
Streams & Patterns
Understands departments and shift types (AM, PM, Night, long day, etc.) as defined in your organisation.
Minimum Rest Periods
Configurable rest rules (e.g. 10h between shifts; longer after nights). Hard rules the AI never overrides.
Maximum Consecutive Days
Enforces limits like “no more than 7 consecutive days” — configurable per organisation and stream.
Leave Integration
Approved leave blocks assignment. The AI never assigns someone who’s on leave — this is a hard constraint.
Fairness Across 7 Dimensions
We track fairness so it’s part of how rosters are built — not an afterthought.
The Future of Roster AI
And we’re nowhere near finished. The more you use it, the more it fits your patterns, your rules, and your language.
Predictive Coverage Modelling
Forecast coverage needs before gaps appear.
What-If Scenarios
Explore roster options before committing.
Richer Fairness Simulations
Forward-looking fairness projections.
Stronger Conflict Resolution
Smarter suggestions when hard constraints clash.
The Result?
A system that is both deeply intelligent and instantly responsive — so departments get the best of both: thoughtful roster generation and fast, accurate answers when they need them.
Technical Summary — For CTOs & Technical Readers
+| Level | Purpose | In Practice |
|---|---|---|
| Level 1 | Streams & shift patterns | What shifts can exist (by department and pattern). |
| Level 2 | Vacancy planning | How many staff are needed per shift, by craft group and scope. |
| Level 3 | Master roster | Repeatable templates and typical patterns. |
| Level 4 | Staff availability | When people can and prefer to work (“My Availability”). Highest priority. |
| Level 5 | The Living Roster | Final assignments: who works when. |
Hard Rules
Rest periods, consecutive days, leave blocks — enforced absolutely. The AI does not override them. Conflicts are visible and auditable.
Soft Preferences
“Prefer mornings” style preferences are optimised when possible, clearly separated from hard constraints in the system.
Different actions consume different credit amounts (e.g. full roster generation vs. a simple query). Usage is tracked per organisation; failed AI calls can trigger automatic refunds. Rate limits are applied per operation type to keep the system stable and predictable.