Skip to content
Neon logo
Deliveries & forecasting

Most modules record.
These two decide.

One works out the order a driver should actually drive in. The other works out the day you have to have ordered by. Both of them run on the records every other module is already writing.

Deliveries Alpha
Forecasting Alpha

Available in Texas only today

Deliveries

Your own routing engine,
not somebody else's API.

The engine is ours, not a metered call out to somebody else's map service. There is no per-request bill, no monthly quota to blow through in December, and no list of your customers' addresses handed to a third party to come back as a line on a map.

How a run gets made

OSRM is the open-source engine that holds the road network. It is built from an OpenStreetMap extract, and what it answers is driving time from any stop to any other over actual roads — not how far apart two pins are in a straight line. We run our own instance of it. The layer above it, the one that decides which driver takes which stop, in which order and in which vehicle, is ours, and it is what the rest of this section is about.

Where it runs

Available in Texas only today. The engine plans against a Texas OpenStreetMap extract that we host, so a second region is a second extract and a rebuild rather than a second product.

01

The drivers you have today

Who is working, when their shift starts and ends, where they set out from and how long a stop takes them. Not an average driver on an average day.

02

The vehicles you have today

What each one can carry and how much of it is already spoken for. A load the van cannot take is not a load the van gets offered, and nobody has to remember that it shouldn't be.

03

The windows you already promised

The slot the customer was given goes into the solve as one of the things it has to satisfy, rather than into the order as a note somebody has to notice.

Tuesday · solved 06:40, re-solved 10:21 Promised window Planned arrival Binding window
07:00 09:00 11:00 13:00 15:00 17:00
M. Alvarez Transit 350 · 1,800 lb shift 07:00–15:00
Round Rock 18 min from depot
Pflugerville 22 min drive
Georgetown 26 min drive
Hutto 21 min drive
D. Whitcomb F-450 flatbed · 5,200 lb shift 08:00–17:00
Cedar Park 31 min from depot
Leander 17 min drive
2,900 lb · flatbed only
Marble Falls 48 min drive
added 10:20
Burnet 29 min drive
Tuesday's planned deliveries, by driver
Driver Vehicle Stop Drive to this stop Promised window Planned arrival
M. Alvarez (shift 07:00 to 15:00) Transit 350 · 1,800 lb Round Rock 18 minutes from the depot 08:00 to 10:00 08:20
M. Alvarez (shift 07:00 to 15:00) Transit 350 · 1,800 lb Pflugerville 22 minutes from the previous stop 09:00 to 12:00 09:22
M. Alvarez (shift 07:00 to 15:00) Transit 350 · 1,800 lb Georgetown 26 minutes from the previous stop 11:00 to 11:30, the binding window 11:12
M. Alvarez (shift 07:00 to 15:00) Transit 350 · 1,800 lb Hutto 21 minutes from the previous stop 09:00 to 15:00 12:20
D. Whitcomb (shift 08:00 to 17:00) F-450 flatbed · 5,200 lb Cedar Park 31 minutes from the depot 09:00 to 13:00 09:40
D. Whitcomb (shift 08:00 to 17:00) F-450 flatbed · 5,200 lb Leander (2,900 lb · flatbed only) 17 minutes from the previous stop 11:00 to 12:00 11:15
D. Whitcomb (shift 08:00 to 17:00) F-450 flatbed · 5,200 lb Marble Falls, added to the board at 10:20 48 minutes from the previous stop 12:00 to 14:00 13:05
D. Whitcomb (shift 08:00 to 17:00) F-450 flatbed · 5,200 lb Burnet 29 minutes from the previous stop 14:00 to 16:00 14:20
Georgetown's 30-minute window is the fixed point that run was ordered around. Hutto's window opened at 09:00, hours earlier, and it is still served after Georgetown — the narrow slot got the part of the day it needed and everything else was arranged behind it. Marble Falls was not on the board at 06:40. It arrived at 10:20, by which time that driver was already between stops, and the run was solved again a minute later — around a van that was out, a shift that was half spent, and every window still owed to somebody. The minutes in the left column come off the road network, not off a straight line drawn between two pins.
Forecasting

A model of your business.
Trained on your business.

Not a template with your logo on it. It learns the shape of your year out of the records your own modules have been writing all along, and out of whatever else you decide belongs in there.

What it learns from

Sales, rental contracts, appointments, stock counts and delivery volume are already in your instance, so the model starts with them and nobody uploads a thing. What it reads beyond that is yours to decide: the weather, the events calendar that empties your town for a weekend, the term dates that fill it again. If your business knows a number moves your numbers, that number can go in — and come back out again when it turns out not to.

What it hands back

A projection, the band it is working inside, and a date. It does not decide what to buy and it does not place the order — you do, and the projection shows you how wide the gap it is guessing across has got.

What this model reads History available Switched off
From the records you already keep No setup. Another module is already writing these.
Sales orders Storefront
104 wk Reading
Rental contracts Rentals
104 wk Reading
Stock counts Inventory
104 wk Reading
Appointments Bookings
96 wk Reading
Delivery volume Deliveries
38 wk Reading
Added by this business Not on any list we shipped. Added because this business knows they move its numbers.
Daily high temperature Added by you
104 wk Reading
Local event calendar Added by you
62 wk Reading
Campus term dates Added by you
40 wk Reading
Diesel price Added by you
12 wk Off
Signals available to this business's forecasting model
SignalSourceWeeks of historyRead by the model
Sales orders Storefront 104 Yes
Rental contracts Rentals 104 Yes
Stock counts Inventory 104 Yes
Appointments Bookings 96 Yes
Delivery volume Deliveries 38 Yes
Daily high temperature Added by this business 104 Yes
Local event calendar Added by this business 62 Yes
Campus term dates Added by this business 40 Yes
Diesel price Added by this business 12 No, switched off
Nine signals. The five at the top were already being written by other modules, so nothing was exported or joined to put them in front of the model. The four below them are this business's own, brought in because somebody there believes they move its numbers — and one of those is switched off again, because they decided it was noise. That is not a decision the software gets to make for them.
The shape it finds

A year has a shape, and it is rarely the one on the wall planner. This is the one the records describe — what the business actually took, month by month, against its own average — and the useful half of it is not the peak everybody already knows about.

Completed rental contracts · two years Above average Below Season nobody declared
1.0× 0
JanFebMarAprMayJunJulAugSepOctNovDec
Formal season Nobody's season Festivals Parties
Demand by month as a multiple of the year's average
MonthContracts completed over two yearsMultiple of the average monthSeason
January 468 0.72 None
February 429 0.66 None
March 663 1.02 Formal season, declared
April 910 1.40 Formal season, declared
May 968 1.49 Formal season, declared
June 715 1.10 Formal season, declared
July 286 0.44 Nobody's season, derived rather than declared
August 332 0.51 Nobody's season, derived rather than declared
September 832 1.28 Festivals, declared
October 845 1.30 Festivals, declared
November 559 0.86 None
December 793 1.22 Parties, declared
Average month 650 1.00 The baseline every month is read against
May runs 3.4× the trade of July, and this business declared three of the four stretches below — the ones it plans for. July and August came out of the records on their own. Nobody writes down the season where nothing happens, which is exactly the season that decides whether June's order is still on the shelf in September.

And then it hands you
a week to order in.

Not a chart to interpret. A list in the order you have to act on it, a day rate the season points at, and the one thing a forecast cannot fix — a supplier slower than the threshold that was set for it.

Every SKU · W31–W40 · today W35 Counted Projected Reorder point
01
Cable Ramp 4ch NN-4185
overdue Order due
02
Truss Section 10ft NN-2277
by W36 Order due
03
Fog Machine 1500W NN-1042
by W38 Order due
04
LED Par Can NN-3310
Clear
05
Speaker Stand Pair NN-5023
Clear
Every SKU's stock cover, ordered by how soon an order has to be placed
Rank SKU Item Counted, W31 to W35 Projected, W36 to W40 Run rate a week Reorder point Weeks of cover Order by
1 NN-4185 Cable Ramp 4ch 72, 60, 49, 34, 22 14, 6, 0, 0, 0 8 26 None, already at or under the reorder point Overdue
2 NN-2277 Truss Section 10ft 92, 78, 66, 51, 38 27, 16, 5, 0, 0 11 30 0 W36
3 NN-1042 Fog Machine 1500W 148, 129, 112, 96, 84 70, 56, 42, 28, 14 14 45 2 W38
4 NN-3310 LED Par Can 318, 290, 268, 240, 210 188, 166, 144, 122, 100 22 60 Does not reach the reorder point before W40 Nothing to place in this window
5 NN-5023 Speaker Stand Pair 138, 127, 119, 106, 96 85, 74, 63, 52, 41 11 30 Does not reach the reorder point before W40 Nothing to place in this window
Three of five need an order placed inside the next five weeks, and one of those is already past the point of placing it. The other two are drawn anyway, because a list that only shows you the problems is a list you still have to check.
Fog Machine 1500W · day rate Charging now Suggested
$40 $56 $64 $48 $57 at W40
W35 W37 W43
Suggested day rate by week, against the rate currently charged
WeekSuggested day rateDifference from the current rate
W35 $48 No change
W36 $48 No change
W37 $51 $3 more
W38 $54 $6 more
W39 $56 $8 more
W40 $57 $9 more
W41 $57 $9 more
W42 $55 $7 more
W43 $51 $3 more
Currently charged $48 Set by you
$9 a day more at W40, starting from W37. Two things point at it and both are yours to check: this line has been out 19 of the last 21 days, and October runs 1.3× the trade of an average month in this business's own two years — the same 1.3× the seasonal figure above draws. It is a suggestion. Nothing here changes a price, and the panel shows its working so you can disagree with it.
Supply · threshold against lead time Cover the threshold buys Supplier lead time
Cable Ramp 4ch Cable factor, overseas · 28d
−0.8 wk Too tight
Truss Section 10ft Rigging fabricator · 21d
−0.3 wk Too tight
LED Par Can Lighting importer · 10d
+1.3 wk Covered
Speaker Stand Pair Audio distributor · 9d
+1.4 wk Covered
Fog Machine 1500W Stage effects wholesaler · 12d
+1.5 wk Covered
What each reorder point buys, against how long its supplier takes
Item Supplier Reorder point Run rate a week Weeks of cover the threshold buys Supplier lead time in weeks Margin
Cable Ramp 4ch Cable factor, overseas, 28 days 26 8 3.3 4.0 0.8 weeks short
Truss Section 10ft Rigging fabricator, 21 days 30 11 2.7 3.0 0.3 weeks short
LED Par Can Lighting importer, 10 days 60 22 2.7 1.4 1.3 weeks spare
Speaker Stand Pair Audio distributor, 9 days 30 11 2.7 1.3 1.4 weeks spare
Fog Machine 1500W Stage effects wholesaler, 12 days 45 14 3.2 1.7 1.5 weeks spare
Two of five reorder points are set below what the supplier actually takes — Cable Ramp 4ch and Truss Section 10ft. Raise the threshold or change the supplier. A forecast can tell you the order is due; it cannot make the lorry go faster, and a figure that implied otherwise would be selling you something we do not have.
Any line on it opens

The same SKU, in full: what was counted, what is projected, the width of the gap the projection is guessing across, and the week it crosses the line you set once. The stock panel above is this, run for everything you stock.

Fog Machine 1500W · units on hand Counted Projected Reorder point
0 80 160 45 Order by W38
W31 Today W40
Fog Machine 1500W, units on hand by week
WeekUnits on handSource
W31 148 Counted
W32 129 Counted
W33 112 Counted
W34 96 Counted
W35 84 Counted
W36 70 Projected
W37 56 Projected
W38 42 Projected
W39 28 Projected
W40 14 Projected
Reorder point45Set by you
Two weeks of cover left at the current run rate, so the order has to be placed by W38. That date is the output. Nobody read a chart to find it, and the run rate it came from is the same sales, rental and booking history the rest of the platform is already writing.
Together

The forecast is what
loads the truck.

Two modules, one set of records, and no step in between where somebody exports a file so the next thing can read it.

01

The forecast names the date

Demand per SKU, weeks of cover, and the day an order has to be placed to land before the shelf empties.

02

The stock arrives against it

Inventory takes the receipt, and the orders waiting on it stop waiting. Nothing was re-keyed to make that happen.

03

The router puts it on a van

Those orders become stops, and the stops become a run against today's drivers, vehicles and windows.

Every module here toggles on and off from the panel, so neither of these is something you have to take to get the other. See what else is in the platform, or ask us what your own operation would look like on it.

Where this stands

Alpha, and specific about it.

Four things worth knowing before you put a week of deliveries or a quarter of stock through either of them.

01

Alpha means early, not absent

Both are switched on for real accounts today, and both count among the 10 modules that are live rather than in development. Alpha here means the surface is still moving. It does not mean a waiting list.

02

Texas today, and that is the map

Available in Texas only today. The engine plans against a map we host, and a map covers what it covers — a limitation of an extract, not a decision about who this is for.

03

Both of them run in your instance

Not a sandbox, and not a trial environment to migrate off later. They sit beside the modules you already use, behind the same isolation, reading the same records.

04

Neither one holds a plan you have stopped agreeing with

Add a signal to the model or take one away, change a driver's shift or move a promised window, and the next answer is worked out against what is true now rather than what was true when somebody pressed go.

Tell us what you run

Both of these are in alpha and both are set up with you rather than handed over. Say what you deliver and what you stock, and we'll tell you what it looks like on Neon.

alpha · configured per business · accounts opened by us