Crisphive

Field service dispatch & route optimization

Agentic AI scheduling, with a deterministic heart

At the center of Crisphive is one deterministic solver — the central scheduler your whole operation runs on. Agentic AI handles the conversation at the edges; data streams, human input and hard constraints feed the core; feasible schedules, optimized routes and cascades feed out. Same inputs, same plan, every time.

  • Feed in — trade requests, live data streams, human-in-the-loop decisions, hard constraints
  • Solve — one deterministic pass: match, route and cascade computed together
  • Feed out — routed days, deterministic cascades, notifications, audit trails
Get started for freeTry the cascade demo ↓

No credit card required

The architecture — Copy link to this sectionLink copied

The deterministic solver is the heart of the system

Everything else — the AI layer, the machine and device layer, your FSM and booking systems — lives outside the core. They feed it and consume from it, but the schedule itself is computed by one deterministic engine. Explainable, auditable, never a guess.

Feeding in

Trade requests

Jobs, service windows, addresses — structured by the agentic AI intake layer.

Human input

Dispatcher pins, overrides and approvals — human-in-the-loop as a hard constraint.

Live data streams

Van locations, telematics, job status and delays from the machine layer.

Hard constraints

Skills, certifications, breaks, supply dependencies, access windows.

EXTERNAL LAYERSCRISPHIVE BOUNDARYINOUTDETERMINISTICSOLVERCENTRAL SCHEDULERAI layerMachine layerFSM / dispatchAPI & MCP

External layers interface with the core — they never place jobs on the board. Indigo streams feed in; orange streams feed out.

Feeding out

Routed schedules

Complete days — travel, mobilization, breaks, supply runs, demobilization.

Deterministic cascades

Emergencies re-flow every affected job in seconds, dependencies intact.

Notifications

Customers and crews informed automatically the moment the plan changes.

Audit trails

Every placement traces to the constraints that forced it — reproducible history.

External AI layer

Agentic AI handles intake, conversation and explanation at the edges — translating messy human requests into structured constraints. It never decides the schedule.

External machine layer

Vans, telematics, IoT sensors and devices stream live signals — location, job status, delays — into the solver as fresh inputs, and receive routed instructions back.

Human-in-the-loop

Dispatchers pin jobs, approve cascades and override decisions. Human input is a first-class constraint — the solver re-plans around it deterministically.

Step 1 / Match — Copy link to this sectionLink copied

Every request meets the one crew that can actually do it

The solver intersects what the customer needs with what your business has on the ground — skills, availability, and who's already nearby. Step through it:

The request — customer side

Job typeAC installation
Skills requiredAC InstallationAC Repair
Estimated on-site120 min
Mobilization / demob15 + 15 min
Service address100 King St W, Toronto
Customer windowsTue eve · Wed AM · Thu AM

Your operation — business side

Priya Patel

Lead · HVAC Technician

FREE WED AM

Emma Smith

Buddy · HVAC Technician

FREE WED AM

James Carter

Lead · Refrigeration

BOOKED ALL WEEK
Van 3 — last job Tuesday · 2.1 km away

Offered to the customer — instantly

JUL14Evening (4PM – 8PM)JUL15Morning (8AM – 1PM)JUL16Morning (8AM – 1PM)

Only windows where skills, people, vans and travel time all line up are ever offered — so a confirmed booking is a booking your business can actually keep.

Steps 2 & 3 / Route, then cascade — Copy link to this sectionLink copied

The whole day, planned like a supply chain

Each booking carries its real anatomy — mobilization, wrench time, demobilization — and the day is routed with travel, protected breaks, supply runs and dependencies built in. Shift one link and everything downstream shifts with it. Then try the button.

Crew A
PRIYA + EMMA · VAN 3
Furnace repairTravel⚡ No-heat emergencyTravelAC install — 100 King St WBreakParts pickupTravelCompressor swapTravelWater heater
Crew B
NOAH + FATIMA · VAN 1
Rooftop unit serviceTravelDuct retrofitTravelWater heater · reassigned
A no-heat call comes in at 10:15. Watch the rest of the day re-plan itself.
Emergency slotted at 10:153 jobs cascaded · dependencies kept1 job reassigned to Crew B5 customers notified automatically

Jobs have anatomy, not just slots

Mobilization, on-site duration and demobilization are planned separately — so a “2-hour job” never quietly eats 3 hours of the day.

Dependencies are respected

Travel, mobilization, demobilization, breaks and supply runs move as one chain — parts stay ahead of the jobs that need them, and buddy crews stay paired, even when the day gets rearranged.

Cascades are deterministic

Re-planning runs on a deterministic solver, not a guess — the same inputs always produce the same feasible schedule, in seconds. Cascades and reschedules are never billed.

Why it matters — Copy link to this sectionLink copied

Bad scheduling is a silent payroll leak

Most field service businesses don't lose money on one dramatic mistake. They lose it an hour at a time — in windshield time, empty driveways and days that fall apart by 10 a.m.

60–65%Typical technician utilization at the average field service organization — a third of every paid hour isn't billable.
2–3 hrsProductive time lost per technician, per day, to scheduling inefficiency. For a 20-tech team that's roughly US$1M a year in wasted labour.
US$1,000+Fully loaded cost of a single missed appointment once rebooking, disruption and lost revenue are counted.
20–30%Utilization improvement reported by companies moving to AI-driven scheduling and routing. That's the gap Crisphive exists to close.

Without a deterministic scheduler

  • One unnecessary hour of driving per day is ~US$39,000/year in lost billable capacity per technician.
  • 47% of appointments don't go as planned — and every disruption is re-juggled by hand.
  • Poor scheduling can put up to 30% of service revenue at risk once missed SLAs, overtime and churn are counted.

With the solver at the heart

  • Routes are optimized around wrench time, not windshield time — travel, mob and demob are planned, not absorbed.
  • Emergencies trigger a cascade, not a scramble — the rest of the day re-plans and every customer is notified in seconds.
  • Only feasible windows are ever offered, so confirmed bookings stay kept — protecting the revenue you already won.

Industry figures: Technology Services Industry Association utilization research; ERPLite & Breaking AC field-service cost analyses (2026); Rev.io MSP dispatch research; Bella FSM State of Field Service 2026; Fieldservicely scheduling research. Illustrative math assumes a US$150/hr billable rate and US$75/hr fully loaded labour cost.

The same solver — match, route, cascade — composed into very different businesses.

HVAC & Mechanical

Residential HVAC contractor

GREATER TORONTO · 14 TECHNICIANS · 8 VANS

Winter no-heat calls used to blow up the dispatcher's whole afternoon. Now emergencies cascade automatically: install crews shift, maintenance jobs reassign, and parts pickups stay locked ahead of the jobs that need them.

+22%

jobs completed per tech-day

−41%

dispatcher time spent re-juggling

HVAC & trades dispatch software →
Last-Mile Delivery

Appliance delivery & install fleet

OTTAWA–GATINEAU · 22 CREWS · 2-PERSON BUDDY TEAMS

Two-person crews, stairs, haul-away and hookup times all vary by stop. Crisphive plans each delivery's real anatomy — load, drive, install, demob — and keeps buddy pairs together when routes rebalance mid-day.

−18%

kilometres driven per route

96%

two-hour windows hit on time

Last-mile delivery scheduling →
Facilities Maintenance

Multi-site property maintenance

NATIONAL PORTFOLIO · 240 BUILDINGS · SLA-BOUND

Burst-pipe and access-control emergencies carry contractual SLAs. When one lands, Crisphive cascades preventive-maintenance work down the priority ladder and proves — per work order — that the SLA clock was met.

99.2%

emergency SLAs met

−31%

overtime hours paid

Facilities maintenance scheduling →
Plumbing & Drain

24/7 plumbing service company

VANCOUVER · 11 TECHNICIANS · ON-CALL ROTATION

Burst pipes don't book appointments. Emergency calls slot into live routes by proximity and skill, after-hours rotations stay balanced, and tomorrow's booked work cascades cleanly instead of getting bumped.

−34%

emergency response time

+17%

billable hours per tech

Plumbing dispatch software →
Electrical Contractors

Service + projects electrical shop

CALGARY · 18 ELECTRICIANS · PERMITS & INSPECTIONS

Panel upgrades wait on permits; inspections come with fixed windows. The solver treats both as hard constraints, weaving service calls around project crews without ever double-booking a licensed lead.

+15%

jobs completed per day

−27%

overtime hours paid

Electrical dispatch scheduling →
Solar & Renewables

Residential solar install crews

MONTREAL · 9 CREWS · WEATHER-DEPENDENT

A rained-out roof day used to mean a week of phone calls. Weather holds trigger a cascade: installs shift, site surveys and electrical rough-ins fill the gap, and every homeowner gets a new confirmed window.

−12

days average install backlog

92%

rebooked same week

Solar installation scheduling →
Telecom & ISP

Fibre install & repair field ops

ONTARIO-WIDE · 60 TECHS · 2-HOUR WINDOWS

Missed install windows are churn. Only feasible two-hour windows are ever offered, outage tickets cascade ahead of installs by severity, and the solver proves per-ticket that response SLAs were met.

97%

install windows hit

−22%

truck rolls per ticket

Telecom field ops scheduling →
Pest Control

Recurring-route pest control

REGIONAL · 260 RECURRING CONTRACTS

Recurring visits, seasonal surges and callback guarantees compete for the same trucks. The solver keeps contract cadences intact while slotting urgent callbacks by proximity — without breaking next month's routes.

+24%

stops per route-day

99%

contract visits on cadence

FAQ — Copy link to this sectionLink copied

How the solution works — quick answers

How does Crisphive's AI scheduling work?
Crisphive works in three stages. Match: every request is matched to the crew with the right skills, certifications, location and availability. Route: the full day is planned with travel, mobilization, demobilization, protected breaks and supply runs built in. Cascade: when an emergency lands or a job runs long, the affected schedule re-plans itself deterministically in under 3 seconds — same inputs, same plan.
What's the difference between route optimization and cascade rescheduling?
Route optimization plans a day once. Cascade rescheduling keeps that plan alive: when reality changes mid-day, Crisphive recomputes every affected downstream job — travel legs, supply pickups, breaks and dependencies included — instead of leaving your dispatcher to re-juggle the board by hand.
Is the schedule decided by an LLM?
The core is a deterministic solver: the same inputs always produce the same plan, which makes every schedule explainable and audit-ready. Agentic AI sits only at the intake and explanation edges — outside the boundary — and it never places jobs on the board.
What does “deterministic” buy me in practice?
Reproducibility and trust. Every cascade can be replayed and explained: auditors get granular history, dispatchers get a plan that never violates a hard constraint, and the same emergency always produces the same feasible re-plan — in under 3 seconds.
Which constraints does Crisphive treat as hard rules?
Skills and certifications, travel time between stops, mobilization and demobilization, protected crew breaks, supply pickups that must happen before the jobs that need them, and access or compliance windows. They are hard constraints, not suggestions — a plan that violates one is never offered.
Does Crisphive replace my existing dispatch or FSM software?
No. Crisphive is scheduling infrastructure, not a walled garden — it runs alongside or inside the dispatch, FSM or booking software your business already uses, through a REST API and an MCP server. You keep your system of record; Crisphive does the scheduling math.
How can I try the scheduling engine before committing?
Use the free sandbox — no signup required. A test-mode API key gives you an isolated test environment against the same base URL, so you can model a real day of jobs and watch a cascade before putting anything into production.

Wire the solver into your product this afternoon

REST API + MCP · 1,000 live jobs/month free · no credit card required

Get started for free