Giving Dispatch Decisions a Consistent, Auditable System

Customer context

VertexOps coordinates field operations across a regional logistics network, historically managed through a mix of phone calls, spreadsheets, and individual dispatchers' judgment and experience.

The challenge

Dispatch quality varied significantly depending on which dispatcher was on shift, and coordination gaps were common during shift changes or high-volume periods when context didn't transfer cleanly between people. There was also no systematic record of why a given routing decision was made, which made it hard to improve the process over time.

Requirements

• Consistent standard-case handling

  Routine dispatch decisions needed to be handled the same way regardless of who was on shift.

• Preserved dispatcher judgment for exceptions

  The system needed to augment, not replace, dispatcher decision-making for genuinely ambiguous cases.

• Full decision audit trail

  Every decision, automated or manual, needed a recorded rationale.

• Real-time fleet awareness

  Routing decisions needed current fleet and capacity data, not a stale snapshot.

Discovery

We shadowed VertexOps' dispatch team for several shifts before designing anything, cataloguing which decisions were genuinely routine — the majority — versus which required real judgment calls informed by context that wasn't written down anywhere. That distinction became the core design principle: automate the routine majority explicitly, and make the exception path better-informed rather than trying to automate it away.

Solution

We built a rules-based routing engine that evaluates incoming jobs against defined criteria — location, capacity, priority — and assigns routine jobs automatically. Genuinely ambiguous cases are queued for dispatcher review with relevant context pre-surfaced, rather than requiring the dispatcher to gather that context manually as before.

Technical architecture

The rules engine evaluates each incoming job against current fleet status, pulled in real time from VertexOps' existing fleet tracking system, and defined routing criteria. Jobs meeting clear automatic-routing criteria are assigned directly; everything else enters a review queue with the relevant fleet, location, and priority context already assembled. Every decision — automated or dispatcher-made — writes to an audit log capturing the criteria or reasoning involved.

Implementation approach

We ran the rules engine in shadow mode initially — generating routing recommendations alongside the existing manual process without acting on them — so VertexOps' operations lead could compare the engine's decisions against actual dispatcher decisions before any automation went live. Discrepancies were reviewed together to refine the routing criteria before automatic assignment was turned on for any job category.

Key features

• Rules-based automatic routing

  Standard dispatch decisions are handled automatically based on defined criteria.

• Exception queue with context

  Ambiguous cases are flagged for dispatchers with relevant context pre-surfaced.

• Full decision audit trail

  Every routing decision is logged with its rationale.

• Real-time fleet status integration

  Routing decisions are based on current, not stale, fleet data.

Integrations

• Fleet tracking system

  Real-time vehicle and capacity status feeds routing decisions.

• Dispatcher communication tools

  Notifications and assignments delivered through existing channels.

Challenges & decisions

The shadow-mode period ran longer than originally planned — several weeks rather than several days — because the first version of the routing criteria disagreed with experienced dispatchers more often than expected. Extending shadow mode to properly refine the criteria, rather than launching on the original schedule, was the right trade-off: it meant the automated decisions were trustworthy from day one of going live.

Representative outcomes

Technology stack

In their words

Related services