Drop in tomorrow's jobs.
stops.csv (id, lat, lng, time-window, service minutes, demand) plus technicians.csv (id, shift, skills, capacity). The same canonical schemas the API consumes.
The pipeline
Seven honest steps from brief to dispatch.
Every step maps to a real surface in the dashboard. No theatre, no fake screenshots.
Catch missing constraints early.
Required-column check, time-window sanity, capacity vs demand, skill coverage. The intake refuses to run a malformed solve so the engine never wastes a slot.
Routes with time windows, capacity, and skills.
The decision engine plans routes against the constraints you uploaded. Plan-aware time budget. Deterministic fallback when an objective is infeasible.
Five scenarios, one decision.
Balanced · Lowest travel · Least overtime · Max jobs · Priority-first. Each rerun is its own job; promote the one your operators agree to.
Route sheets a driver can read.
PDF Operations Plan with route per technician, ETAs, and service-time blocks. DOCX brief for the morning standup. JSON for your dispatcher.
Catch drift before it cascades.
When a job runs long, an alert surfaces with the affected stops and an Apply-fix path that re-sequences the rest of the day.
Webhooks back into your dispatch tool.
solve.complete, decision.approved, alert.threshold — every event is HMAC-signed, idempotent, and traceable to a request id.
Example output
What tomorrow's plan actually looks like.
Every preview below is generated by the live engine on real demo data — visible in the dashboard the moment your team signs in.
Trade-offs at a glance
Same problem. Three honest objectives. Promote the one your operators agree to.
RecommendedBalanced
score 86
Travel0 min
WorkloadEven
Overtime0 min
Lowest travel
score 74
Travel0 min
WorkloadUneven
Overtime0 min
Least overtime
score 78
Travel0 min
WorkloadEven
Overtime0 min
Pulse stream
solve.completejob_jK7Pq2 · request_id req_xVxN…6s ago
alert.thresholddepot.NJ p95 routes drift > 2σ1m ago
decision.approvedops@ approved Balanced plan4m ago
Two stops in James's afternoon route are tracking 14 minutes behind window. Apply the alternate sequence?
Apply fixCompareIgnore
Eight formats. One narrative cache.
Real PDF / DOCX / PPTX / XLSX generated server-side, with branded covers and a narrative cache so repeated exports are sub-second.
PDFOperations plan
DOCXBrief
PPTXDecision deck
XLSXScenario book
HTMLStandalone report
CSVRaw KPIs
JSONResult payload
ZIPBundle
Plan the next 24 hours in 15 minutes