Deterministic & reproducible
Order-independent. The same stops always yield the same plan — diff-able, auditable, SLA-ready.
// same input → same route
Deterministic route-optimization engine
One optimized route.
O(n log n) scaling, not a budget
One optimized route.
Every stop. Every run.
iRoutify plans up to 20 million stops in minutes — one near-optimal, crossing-free route, identical on every run. Not a time budget. Not per driver.
20Mstops · ~9.7 min
100%deterministic
0route crossings
The problem every fleet inherits
Every other tool makes you pick a side.
01 · SPEED vs QUALITY
Fast or optimal — never both
Heuristic apps answer in seconds with approximate routes; OR-grade solvers run for hours or a fixed time budget. iRoutify returns one deterministic, near-optimal answer in linearithmic time.
02 · SCALE CEILING
The wall at a few thousand stops
Per-seat planners cap around 1,000 stops and are split by hand beyond that. iRoutify solves the whole network — up to 20M stops — as one problem.
03 · REPRODUCIBILITY
A different plan every run
Metaheuristics drift run-to-run, so plans can't be audited or diffed. iRoutify is deterministic and order-independent: same input, same plan, always.
What you get
A routing core built for the hard requirements.
The properties enterprises actually procure on — reproducibility, guaranteed quality, web-scale throughput, data residency — are the defaults, not the roadmap.
Checked crossing-free
Every route whose order only distance decides is checked for crossings before it is returned, and returned without one — the "spaghetti" artifact that erodes driver trust and inflates miles never ships.
// 0 self-intersections, checked on every response
Linearithmic at 20M stops
Planning time grows in step with the number of stops (n log n), whatever their layout: evenly spread, clustered, along a single road, with far outliers or with many stops at one address. Measured on eleven deliberately difficult layouts up to one million stops; in every one the work stayed inside the same fixed multiple of n log n. 20 million stops in ~9.7 minutes on one 8-core box, using 5.6 GB — measured end-to-end, not extrapolated.
// 20M stops · 9.69 min · 5.6 GB · measured
Self-hosted or Azure SaaS
Run it inside your own cloud or on-prem — data stays in-jurisdiction, with no hosted quota — or consume it as managed Azure SaaS.
// on-prem · any cloud · SaaS
Native fleet & time-window split
Balance one job across N vehicles or pack routes into time budgets in a single call — no external splitter, balanced across vehicles.
// N-vehicle · time-boxed
Flat-rate & open REST API
Priced flat, not per driver — growth never inflates the bill. An open REST API drops the engine straight into your stack, OEM-ready.
// flat-rate · embed in weeks
The requirements real fleets carry
One solve, the whole constraint model.
Windows, capacity, driver hours and depots are one request to iRoutify — not a pipeline of external splitters. Every one is live in the API today.
Delivery time windows
Several windows per stop, hard or soft, with per-vehicle shift start and end. Late stops are dropped in hard mode, never hidden.
Vehicle capacity
Any named dimensions — pallets, weight, volume — enforced as a running load at every leg, not just the route total.
Pickup & delivery
A pickup and its drop stay on one vehicle, collect-before-deliver, within a maximum carry time. Split a pair and both halves are withdrawn.
Multiple depots & open routes
Many depots in one solve, each vehicle its own start and end, returning to base or finishing in the field.
Driver hours & calendar
Breaks, daily driving and duty limits, plus fixed meetings, floating lunches and an overtime allowance — built into the timetable, not checked after.
Service areas & avoid zones
Restrict a vehicle to its territory, or bar delivery inside a polygon, with every route reporting the legs that clip a zone.
Route caps & unserved reasons
Hard limits on stops, distance, duration and radius per route. Every stop comes back either on a vehicle or in an explicit, reason-coded unserved list.
Capabilities, priority & must-serve
Match stops to vehicles by required skill or equipment, rank by priority, and pin deliveries that can never be dropped.
Stop profiles
Describe a kind of consignee once — its windows, dwell, equipment and load — then point every stop like it at that. Anything a stop states for itself still wins, field by field, and the answer says what each one inherited.
Multi-day horizon
Plan a span of days in one call; the engine chooses each stop’s service day and levels the work across the week.
Workload balancing
Equalise work across the fleet by stop count, distance or load — the fairness trade reported on every plan, balanced or not.
Live-plan insertion
Slot a new order into a plan already being driven, ranked by true added cost, without resequencing the stops already dispatched.
Mid-day re-dispatch
Repair the rest of a day that has changed. Deliveries already made are frozen, each vehicle re-enters the problem from where it actually is, and the answer lists exactly what moved and why.
Quality certificate & replay
Each route reports what was proved and against which bound; a signed digest and a replay endpoint verify same-input, same-plan.
Plan versus actual
Score a finished day against the plan it was handed: stops served out of order, stops missed, extra distance driven, and how much of the lateness the plan had already admitted to. Anything you cannot report is left unknown, never estimated.
How it compares
The tier no one else covers.
Per-seat SaaS, hosted VRP APIs, and captive in-house engines each solve part of the problem. iRoutify is the one engine that is deterministic, crossing-free, web-scale, and yours to run.
| Per-seat SaaS | Hosted VRP API | In-house engine | iRoutify | |
|---|---|---|---|---|
| Max practical scale | ~1,000 stops / plan | 2,000 orders · 100 routes / job | Web-scale | 20M stops / job |
| Solve behavior | Approximate, time-budgeted | Async job, ~4-hour window | Varies; proprietary | Deterministic, near-optimal (O(n log n)) |
| Determinism | — not guaranteed | — not documented | ? undisclosed | ✓ guaranteed |
| Crossing-free routes | — | — | — | ✓ checked on every route |
| Pricing | Per driver / month | Metered per route / credit | $100M+ & a decade to build | Flat-rate |
| Deployment | Vendor cloud | Vendor cloud | Captive, never licensed | Self-host · any cloud · SaaS |
| Time to value | Fast, but capped | Fast, but metered | Years | Integrate in weeks |
Competitor figures are drawn from vendors' own public pricing, docs and limits pages (e.g. hosted-VRP caps of 2,000 orders / 100 routes per job; per-seat planners capping around 1,000 stops). Comparison for evaluation only; trademarks belong to their owners.
By the numbers
Different class of number, not a better setting.
20M
stops per single solve
9.7min
measured, to optimize 20M on one 8-core box
100%
reproducible, run to run
15–30%
fleet opex route-opt can save
Who it's for
Built to sit under the operation.
Fleet operators
Last-mile & field service
Delivery, courier, grocery, inspection and service fleets that have outgrown per-seat planners and need whole-network routes, daily.
Platforms
TMS & logistics software
Embed a deterministic, web-scale optimization core behind your own UX via REST — the tier your roadmap points to, without building it.
Enterprise
On-prem & data-resident
Organizations that can't ship routing data to a vendor cloud — run the whole engine inside your own perimeter, in-jurisdiction.
Try it on your data
Point it at your hardest dataset.
Open the live app, drop in your stops, and watch one deterministic, crossing-free route come back — in seconds.
Alexei Corduneanu · Founder
alexei@iroutify.com