Understanding conflicts

What counts as a conflict, every conflict kind explained, and how conflicts surface across the product.

A conflict is a scheduled request that violates the planning model right now — overlapping bookings, a missing capability, a placement outside availability. Orkyo’s stance: plans drift, so instead of blocking every edit, it continuously detects violations and shows them where you work. “No conflicts detected. All scheduled requests meet their requirements.” is the healthy state.

Why conflicts exist at all

Validation at scheduling time catches problems at placement. But schedules are living things: a capability gets removed, an absence gets recorded, a second request lands on the same slot, a site’s working hours change. Any of these can invalidate a placement that was fine yesterday. Conflict detection is the safety net that keeps the schedule honest after the fact.

The conflict kinds

As shown on the Insights → Conflicts tab:

Kind Meaning Typical fix
Scheduling Overlap Two exclusive bookings collide on the same resource Move one; the card links to “View other request”
Capacity Exceeded A station holds more than its Capacity allows Move work, or correct the capacity if it’s wrong
Load Exceeded A fractional resource’s allocation budget is overspent Rebalance assignments
Capability Mismatch The resource no longer satisfies a requirement Restore the capability, drop the requirement, or move the request
Below Minimum Duration The placed slot is shorter than the request’s minimum Extend the slot
Before Earliest Start / After Latest End The placement breaks the request’s own time constraints Move it inside the window, or adjust the constraint
Size Mismatch The resource’s size doesn’t fit the requirement Pick a fitting resource
Dependency Violation The request starts before the one it waits for finishes, or that predecessor is unscheduled Move the successor later, schedule the predecessor, or remove the dependency

A dependency violation has two severities. A successor that starts too early is an error: the plan is wrong. A successor with too few scheduled predecessors to judge is a warning: the plan is incomplete, and nothing is wrong yet.

A request does not wait for every predecessor in every case. If its start condition is already met, Orkyo reports nothing. See Dependencies and the critical path.

Trend reporting groups these into six categories — Overbooking, Criteria mismatch, Resource unavailable, Outside availability, Missing resource, and Sequence violation — which is what the Insights conflict chart plots over time.

Where conflicts surface

  • On the request — a banner counts “conflicts on this request”, with each conflict’s message.
  • On the board — conflict indicators mark affected placements where you schedule.
  • In Insights → Conflicts — the full list as cards, filterable to a single request, each linking to the counterpart request where one exists.
  • In the trend chart — whether conflicts are growing or shrinking over the period.

Working the list

Conflicts are cheap to detect and cheapest to fix early. The productive loop: scan Insights → Conflicts, fix instances on the Utilization board, and watch the summary line count down. The hands-on version is Find and resolve conflicts.

A recurring conflict is feedback about the model, not just the schedule — a machine that is permanently Capacity Exceeded has a capacity problem, not a scheduling problem.

The AI assistant can also explain a conflict and propose a correction, which you approve or refuse.