An optimiser that builds the roster. Not a grid that shows your mistakes.

The roster is produced by an industrial-grade mathematical optimisation engine (Google OR-Tools CP-SAT), not by a generative chatbot. That matters in healthcare because the result is deterministic and repeatable: the same inputs give the same roster, and hard rules are constraints the engine cannot break. A drag-and-drop grid only shows you where you broke a rule after the fact. Roster searches for the best compliant roster for you. An AI assistant connector for ChatGPT and Claude (via MCP) is on the roadmap and currently in pilot.

Ten built-in hard rules are enforced on every run: no double booking, role eligibility, no outpatient/inpatient cross-deployment unless permitted, no work on approved leave, no work on committed clinic days, minimum rest after night shifts, a maximum number of consecutive working days, a weekly shift cap, minimum headcount for every slot every day, and honouring pre-agreed night allocations. An alternate-weekend rule can be switched on, and every slot also enforces its required skills, in-date credentials and minimum seniority. A library of 18 configurable hard and soft rule types is on the roadmap.

The engine never hides a shortfall and never breaks a hard rule to cover it. Anything it cannot fill appears in the Gap Report with the date, the slot, the required versus assigned headcount, the size of the shortfall and a suggested remedy, so you can arrange agency, overtime or redeployment weeks ahead instead of on the day. Any unfilled slot can also be explained on demand.

The fairness engine looks back over each person's actual duty history (six months by default) and prefers staff who have had fewer recent nights, weekends and duty shifts, so the load evens out over time. The Fairness Summary report shows the distribution of nights, weekends, Sundays and public holidays per person, so fairness is measurable and easy to defend to staff.

Yes. Control stays with the planner. You can add, edit, swap or delete assignments from the calendar, and manual edits are checked for conflicts and for eligibility in the same way as generated ones. Eligibility gaps on manual edits can be set to Hard Block or Advisory. Every assignment records its source (AI proposed, manual, manual override, night-shift import or rebalance) and every change is audited.

Techseria Roster is designed for hospital departments of about 180 staff across multiple sites and roles, and bulk employee import is sized for 50 to 500 employees. The optimiser runs as a background job that can have its own worker queue, so heavy runs do not slow everyday HR and ERP work. One installation can serve several hospitals, polyclinics or pharmacy outlets in a healthcare group.