One operating system for the whole operation.
Predictive Maintenance
Weibull reliability curves and 10,000-run Monte Carlo forecasts tell you what fails, when, and what it costs.
AI Schedule Optimization
A multi-constraint VRP solver assigns the right coach and driver to every booking, FMCSA-legal by construction.
HOS Compliance Gates
Passenger-carrier 49 CFR 395.5 limits are solver dimensions; illegal assignments are blocked before dispatch.
Quote-to-Cash Automation
Quotes become bookings, trips, invoices, and online payments without re-keying a single field.
Live Customer Experience
Tracking links, ETA texts, and online payment replace every "where's my bus?" phone call.
Self-Hosted AI
Your fleet data never leaves your cloud: open-weight models on owned GPUs, zero per-token bills.
Sentinel · Predictive Maintenance
Fix it in the shop. Not on the shoulder.
10,000
Monte Carlo simulations
Every cost forecast runs ten thousand simulations, split into labor and parts, per vehicle and fleet-wide.
30/60/90
Day failure predictions
Maintenance cost estimates and remaining useful life stated with 95% confidence intervals, in days and miles.
VMRS
Coded component reliability
Weibull survival and hazard curves per component, keyed to the fleet-maintenance industry standard.
Sentinel computes remaining useful life in days and miles for every vehicle, ranks component-level risk keyed to VMRS codes, and runs anomaly detection that flags atypical wear long before failure. A critical DVIR defect grounds the coach automatically, before dispatch instead of on the highway.
Schedule Optimization
One click assigns the right bus and the right driver. Legally.
FleetHarmony solves the day's bookings as a full multi-constraint vehicle-routing problem — Google OR-Tools pickup-and-delivery with time windows, plus a mixed-integer solver in the same class of technology as airline crew scheduling. Illegal assignments are filtered out before anything is proposed.
Hours-of-Service law
Passenger-carrier 49 CFR 395.5 limits — 10-hour drive, 15-hour duty, 60/70-hour cycle — are solver dimensions, not after-the-fact warnings.
Driver credentials
CDL validity, medical certificate, passenger endorsement. Expired credentials hard-block the assignment.
Vehicle readiness
Annual inspection, maintenance-due status, active fault codes, and DVIR state gate every vehicle.
Capacity & convoys
Seats matched to group size, with multi-vehicle convoy assignment for oversized groups.
Deadhead economics
Empty miles are penalized 1.5× and chained trips cut them; real road times come from an H3 travel-time matrix.
DVIR buffers
20–40 minute pre- and post-trip inspection blocks are built into every duty envelope, so schedules are executable.
Time windows
Pickup and dropoff windows per booking, solved as a full pickup-and-delivery problem with time windows.
The solver proposes; the dispatcher disposes. Results stream in live as ranked recommendation cards with score tooltips and compliance-violation flags — and nothing is applied until a human reviews and approves it.
Compliance & Safety
FMCSA-first, by construction.
Passenger-carrier rules, not borrowed from trucking
Correct 10-hour drive, 15-hour on-duty, and 8-hour off limits per 49 CFR 395.5 — the rules motorcoach operators are actually held to.
Violations blocked at dispatch, with citations
Compliance gates reject illegal assignments with CFR-referenced violations before a coach rolls — not at a DOT audit.
A rule engine you control
A tenant-customizable decision-table engine ships with 15 seeded FMCSA rules, supports overrides, and keeps a full audit trail.
Critical defects ground the coach automatically
FMCSA-structured DVIR inspections with E-SIGN signatures escalate defect severity; a critical defect takes the vehicle out of service with no dispatcher action required.
Seven years of audit trail
Every data mutation is recorded with before-and-after values and retained for seven years, with GDPR export built in.
Customer Experience
Your customers watch the bus arrive.
Automated messages at every milestone and a live tracking page end the “where's my bus?” phone call — for good.
- Instant public quotes with branded PDFs and online acceptance
- Day-before and departure SMS, live ETA loop, geofence arrival alerts
- No-login trip tracking with driver photo and ETA timeline
- Stripe invoice payment with early-pay discount
- Two-way messaging between customers and dispatch
AI & Data Sovereignty
Your fleet data never leaves your cloud.
The entire AI stack runs on owned GPUs inside your deployment: open-weight models served by vLLM, retrieval-augmented search over a knowledge graph of your operation, and an assistant grounded in live fleet data with visible citations. No per-token bills, no third-party model providers, no data leaving the VPC.
Grounded answers, visible citations
The assistant answers from live fleet data with citations and visible tool calls — and role-based access control filters every tool it can reach.
Guardrailed by default
Read-only natural-language SQL, prompt-injection and PII guardrails on all AI input and output, and AI-drafted emails that are always drafts — never auto-sent.
Natively agent-callable
A built-in MCP server exposes the platform to agentic workflows, with retrieval kept fresh across the operational data model.
Migration
We migrate your history, not just your login.
One command moves a legacy Coach Manager database into FleetHarmony with full fidelity: chart of accounts, general ledger, AR aging, rate cards, and pricing schemes — proven end-to-end at real scale with zero extraction errors. Nothing is silently dropped; quarantined rows are reported.
103,265
trips
26,910
bookings
14,858
invoices
82,745
line items
53,920
GL entries
0
extraction errors
The architecture of fleet operations.
103,265
Trips migrated with zero extraction errors
2 min
To a fully branded demo tenant
44
Durable automation workflows
8
Languages in the driver app
Two minutes from now,
this is your fleet.
Scan, enter your work email, and get a fully branded FleetHarmony tenant with your colors, your locations, and a calibrated demo fleet.