AI dispatch alerts that catch service problems before your riders do
Aptimized Dispatch Assist watches every vehicle in real time, predicts bus bunching, service gaps, and breakdowns minutes ahead, and hands your dispatcher a one-click recommendation. Your people stay in charge. Your current dispatch system stays in place.
Delivered by Aptimized with Samsara connected operations and Google Cloud AI.
Projected to bunch at 42nd St in 8 minutes. Confidence 91%.
Hold bus 4417 at 34th St for 90 seconds.
Bus 4421 is 1.8 minutes late after a signal delay. Holding 4417 restores a 4.8-minute headway with no impact on the trip schedule.
Illustrative example. Try Accept or Override.
At rush hour one dispatcher can be responsible for 40 buses on a single route. No one can watch them all at once.
Dispatch Assist watches every vehicle, every segment, every 30 seconds, and only interrupts your team when something needs a decision. Problems get handled in seconds instead of minutes, and your dispatchers spend their attention where it matters.
What operations managers get
Automatic dispatch alerts and decision support built around the way your control room already works.
Early-warning alerts
Bunching, headway gaps, schedule slips, abnormal dwell, and passenger-load surges are flagged automatically, ranked by urgency, and shown on a route risk map.
One-click recommendations
Hold, leapfrog, short-turn, swap, return to depot, or keep running. Each suggestion explains why in plain English, shows projected impact, and is logged when accepted or overridden.
In-cab guidance
Short, clear prompts on the cab tablet: adjust speed, hold at the next stop, watch the crosswalk ahead. Alerts respect operator workload and never compete with safety.
Predictive vehicle health
Fault codes and engine data reveal buses likely to fail on the road. The system suggests a swap, pre-stages the replacement, and opens the work order automatically.
Shift and network KPIs
Live dashboards for bunching, excess wait time, headway adherence, time-to-intervention, and recommendation acceptance, by route, depot, or the whole network.
Rules you control
Risk thresholds, alert routing, and when to suggest each action are set by your team, per route, depot, or time of day, without a developer or a vendor ticket.
One morning on a busy route
What your dispatcher, operator, and depot see during a single AM peak.
The system spots it first
Two buses are 90 seconds apart when they should be five minutes apart. An alert fires with a projected bunch location and a confidence score, and moves to the top of the dispatcher's queue.
One click restores headway
The dispatcher reads the recommendation to hold the lead bus for 90 seconds and accepts it. If there is road work at that stop, they override with a reason code, and the system learns from it.
The operator gets a clear prompt
The cab tablet says to slow slightly and hold 15 seconds at the next stop, plus a pedestrian alert for a busy crosswalk. One tap to acknowledge.
The bus warns you before it breaks
A coolant fault matches a known failure pattern. Dispatch and the depot get the same alert at the same time, a replacement bus is ready, and the swap happens before a rider ever notices.
Everyone sees how the route performed
Bunches prevented, wait time against baseline, and every recommendation, acceptance, and override, ready for the next shift briefing.
Improve the process you have without adding risk
Dispatch Assist is an optimization layer, not a replacement. It is introduced gradually, measured at every step, and reversible at any point.
Keep your CAD/AVL and dispatch console. Recommendations appear inside the tools your team already uses, with single sign-on. No second screen.
Start with advice only. In the pilot, every action is reviewed by a dispatcher. Nothing is automated until your team decides it should be.
Pilot one depot first. Each rollout phase has a measurable gate and needs your sign-off before it expands.
Fail safe. If any service slows or drops, the system falls back to standard rules and your existing process keeps running.
You own it. Your data, the models trained on it, and your configurations belong to you, with a full export if you ever leave.
Public-sector security. Built on Google Cloud with FedRAMP High authorization and Samsara telematics with a provisional GovRAMP designation, encrypted end to end, with US data residency.
You choose how much the system does
Monitor
Live alerts and telemetry. No recommendations.
Advise
Recommendations issued. A dispatcher reviews every action.
Suggest and execute on click
The dispatcher decides each event with one click.
Auto-execute low-impact actions
For example, rider info updates. High-impact actions still wait for a dispatcher.
Supervised automation
Acts within rules you set. Dispatchers handle exceptions.
Moving up a level is a decision your command center approves, typically only after dispatchers are accepting at least 70% of recommendations.
Built for the people who run the service
Technology in the control room succeeds or fails on trust. Dispatch Assist is designed around your workforce, your labor agreements, and the realities of a 24/7 operation.
Labor and union partnership
A tool that supports represented employees, with no surprises.
- The system recommends. Dispatchers decide. It never overrules a person.
- Operators can respond to in-cab guidance with reason codes, so their judgment on the road is recorded and respected.
- Dash-cam video is available in about 30 seconds, helping clear drivers of false claims instead of leaving them in limbo.
- How safety data is used for coaching is a policy your agency sets. We recommend bringing labor representatives in at the design stage and agreeing on it before launch.
- Every recommendation, acceptance, and override is logged, giving both management and labor a transparent record.
A user experience people adopt
Fewer screens, clearer signals, less noise.
- Alerts live inside the existing dispatch console. No new app to learn for dispatchers.
- Plain-English reasons for every recommendation, not a black box score.
- Accept or override in about a second.
- In-cab prompts are short and timed around route phase and workload.
- Alert false positives are tracked and tuned down, with a design target of 8% or less.
Change management from day one
Adoption is planned and measured, not assumed.
- Readiness assessment with dispatch, operations, maintenance, and labor stakeholders.
- Dispatcher cohort training in person and online, plus operator orientation on the cab tablet.
- Train-the-trainer so your own supervisors own the program.
- Office hours and floor support through the pilot and expansion.
- A 90-day tuning sprint driven by your dispatchers' accept and override decisions.
Measured against the service metrics you report on
The platform is engineered against five service quality targets, baselined in your pilot and tracked weekly.
less on-route bunching
less excess wait time
better headway on high-frequency routes
faster time from problem to intervention
dispatcher acceptance of recommendations
Figures are design targets, not guaranteed outcomes. Results depend on fleet, routes, data quality, and operating conditions and are confirmed against your own baseline. Design latency targets: 20 seconds or less from vehicle event to dispatcher recommendation, 10 seconds or less for in-cab alerts.
A rollout that proves itself before it grows
Discover and design
Control-room walkthroughs, data review, success criteria, and labor and stakeholder alignment.
Single-depot pilot
Telematics installed, models trained on your history, alerts running in advisory mode.
Regional expansion
Scale to a division or region. Enable low-impact automation only if you approve it.
Network and steady state
All depots live, with ongoing tuning, retraining, and quarterly reviews.
Who Dispatch Assist is for
Any operation where a dispatcher manages vehicles against a schedule or headway, in the United States and internationally.
Questions transit leaders ask
Do we have to replace our CAD/AVL or dispatch system?
Will the AI make dispatch decisions on its own?
How does this affect union-represented dispatchers and operators?
What results can a transit agency expect?
How fast are the alerts?
Who owns the data and the AI models?
How long does a pilot take?
Which technology does Dispatch Assist run on?
See it against your own routes
In a two-hour working session we walk your operations team through a live shift scenario and map it to your dispatch system, labor environment, and service targets.
- Walkthrough of alerts, recommendations, and in-cab guidance
- Integration fit with your current CAD/AVL
- Pilot scope, gates, and change-management plan