Constraint optimisation, guided by planners

Automated rostering helps ATC planners compare assignments against demand, qualifications, availability and scheduling rules. Planners set the requirements, inspect the result and decide what is ready to publish.

SkyRoster uses constraint optimisation to construct and improve rosters from configured inputs. Its production engine does not learn from customer rosters or forecast demand through machine learning.

Set the planning requirements

Shift types define working periods. Rotation patterns describe sequences of shifts and days off. Staffing requirements specify the numbers and qualifications needed for a shift or working position.

Planners also need to maintain availability, leave and competency records. When demand exceeds available qualified staff, optimisation can help explore alternatives, but it cannot create additional capacity.

Make room for staff input

Include preferred shifts and days off in the plan. The engine balances those preferences with coverage and other constraints, so some requests may remain unmet.

SkyRoster brings shift-swap requests into an approval workflow. The relevant rule checks and local policy guide the review; replacements are not independently approved by machine learning.

Keep the decision with planners

The engine scores candidate rosters against configured constraints. Planners review the result, investigate violations and decide whether further changes are needed before publication.

Trainee and instructor assignments require particular care: SkyRoster includes checks for these assignments, but planners configure and review the pairings. Instructor pairing is not solved independently by the engine.

Roster versions help planners manage changes and publication. The review process should make clear which version staff should follow and who is authorised to publish it.

Measure the difference

A useful evaluation compares preparation time, revision effort, coverage, overtime and the distribution of duties against the current process. Keep the demand, staffing assumptions and rules consistent between alternatives.

Automation can reduce repeated calculation and data handling. The organisation remains responsible for its rules, staffing decisions and operational review, and should measure the benefits against its own baseline.

Prepare an evaluation

1. Select a representative roster period and anonymise personal details.
2. Document staffing demand, shift patterns, qualifications and rules.
3. Recreate the current plan and check that its inputs are represented correctly.
4. Generate and compare alternatives using agreed measures.
5. Review exceptions, integrations and implementation requirements before deciding how to proceed.

SkyRoster focuses on ATC rostering. Predictive analytics and other proposed AI uses should be evaluated separately from the production rostering engine.

Read how the engine builds and improves a roster, then see how Check Solution supports review of the current assignments.

Updated September 2026: this article now describes the engine as constraint optimisation and removes earlier claims of automatic compliance and machine-learning behaviour.