Lastmily · Platform · Last mile
Last mile · the flagship
From an order file to the last stop on the round.
Import the orders, let the engine sequence them, hand the round to your own vans or to a carrier you buy from, and run it in the driver’s hand – with the recipient watching a branded page the whole way.
End to end
Six steps, one record.
Plan, assign, execute – and one event log underneath the whole thing.
Shopify · API · ERP · WMS · XLS
five models, from single driver to preference‑based
your vans, your drivers
carriers and partners you buy from
stops grouped by entrance and floor · scan · PoD · COD
branded tracking · notifications · change it from a phone · rating
BOX NOW and Clever Point integrated end to end
The apps, where they actually run
Scroll the line sideways to follow the round →
The apps, where they actually run
Shopify · API · ERP · WMS · XLS
five models, from single driver to preference‑based
your vans, your drivers
carriers and partners you buy from
stops grouped by entrance and floor · scan · PoD · COD
branded tracking · notifications · change it from a phone · rating
BOX NOW and Clever Point integrated end to end
- Dispatcherimport · route · assign · re-plan during the day
- Driver appnavigation · scan · PoD photo, signature, PIN · COD, Soft PoS, NFC
- Experience layerbranded tracking · proactive notifications · redirect · NPS
- Networks & PUDOlockers, pick-up points, carrier selection
- Control towerone event log · live map · ETA · KPIs · alerts
Your own vans and the carriers you buy from sit on the same line – one sequence, one event log, one report.
Lastmily · Last mile
What is inside the block
Five products, one round.
Routing engine
Five ways to build a round – single driver, regional, dynamic regions, free, or preference-based free – with one solver underneath. Around twenty seconds to solve 700 stops, so a dispatcher re-plans rather than accepts.
Execution & live monitoring
One event log, a live map, ETA, KPIs and alerts. The day as it actually happens, not as it was planned this morning.
Driver app & PoD
Route, stops, navigation grouped by entrance and floor. Scan, deliver, pick up. PoD photo, signature, PIN. Exceptions and self-debrief.
Settlement & COD
COD, Soft PoS and NFC at the door, reconciled back to the shipment and the shipper without a second system.
Last Mile AI
Preferable time windows, area-point matching, service time prediction and real world routing – reading the same record every block writes to.
Execution & live monitoring
One canvas for the whole day.
The dispatcher works the map or the driver rail, whichever way the problem arrives. A stop opens into its own history – routed, assigned, notified, attempted, delivered – so the answer to “what happened to this parcel” takes one click, not one phone call.
The KPI bar is not a separate reporting product. It is the same event log read at a different cadence, which is why the number on the screen and the number in the month-end report do not disagree.
Lastmily · Last mile
Driver adoption
The route that gets driven is the only one that pays.
Adoption is not a training problem to be solved after go-live. It is the design constraint the whole block is built around – because in B2C, typical route adherence is around 35%: an industry baseline, not a Lastmily result.
Here is what usually happens. A new system arrives. The drivers are given a login and a route that does not match the street they know. At first they follow it. Before long they are driving their old sequence and marking stops in batches at lunchtime. The data goes bad, the ETAs go bad, the recipient stops believing the notification, and the dispatcher goes back to the phone.
Nothing in that chain is a software fault. Every link is an adoption fault – and it compounds, because the model that was supposed to learn from the day is now learning from fiction.
So the block is built the other way round. The app asks for almost no training. The route is planned inside areas the driver already works, in the order he already drives them. He can tell the system where the entrance is and where he wants to finish, and it listens. What he corrects at the door goes back into the model rather than into a support ticket.
Adherence then stops being a stick to beat drivers with and becomes a measurement of whether the plan deserved to be followed.
Day one · almost no training
Scan-first, one-hand operation, one screen per task. A driver who has used a phone can work a round. No classroom, no manual, no super-user.
First rounds · the route looks familiar
Dynamic regions are drawn from your own delivery history, so the areas are the ones drivers already recognise – not lines somebody once drew on a postcode map.
Then · the driver corrects it
“This is the entrance.” “I want to end here today.” Address corrections and access notes feed back into the model instead of being lost in a WhatsApp group.
Later · the model has moved
Service time, preferable windows and area-point matching have absorbed real behaviour. The plan now proposes what the good driver was already doing – which is when the rest of the fleet starts following it too.
Ongoing · adherence is a KPI, not a hope
Route adherence sits in the control tower next to first-attempt success. When it drops on a round, that is a signal about the plan, and it is visible the same day.
Driver · mobile
Built for one hand, in the street, in the rain.
Route, stops and navigation grouped by entrance and floor – so a multi-parcel building is one stop with a plan, not a list of rows to hunt through in a lift.
- Scan, deliver, pick up
- PoD photo, signature, PIN
- COD, Soft PoS, NFC
- Exceptions & self-debrief at the end of the round
The driver closes his own round: exceptions, undelivered, cash. The office stops re-keying a paper sheet, and the record is complete before the van is back.
Being able to say “I want to end here today” is not a courtesy. It is the difference between a tool a driver uses and a tool a driver works around.
An address fixed at the door is worth more than the same fix typed into a CRM days later, because it reaches the next route.
Ask us about adherence, not about optimisation.
We will show you what your drivers are doing with the plan today, and what the territory model does to that number.
Lastmily · Driver adoption
How AI is reshaping the last mile
Four things the engine learns that a textbook cannot.
01 / The driver
Ten days replayed, six areas learned
Then the engine plans inside them, in his order.
02 / The street
Stop at the cross, walk the one-way street
Drive to the next corner – not a lap for every door.
03 / The recipient
When this door actually opens
Inferred from what happened there before.
04 / What is coming
A pickup expected at 11:00
The reactive van drives all the way back; the predicted one is already there.
Lastmily · Last Mile AI
Where it lands
Door, locker or PUDO point.
The drop-off is a routing decision, not an exception. BOX NOW and Clever Point are integrated end to end; ACS, Courier Center, Easy Mail and Speedex are connected – and we will connect anyone you already use.
The door
Planned against the window this door actually opens in, with the entrance and floor already known and the walk planned as well as the drive.
The locker
A first-class stop on the round with its own parcel count – and, for a recipient who collects from the same locker every time, often the better first attempt.
The PUDO point
Yours or a partner’s. Either way it is a node on your network, measured with the same partner-performance model that chooses it.
Keep reading
Lastmily · Last mile
The close
Pick the number we should be measured on.
Then let us measure it before you commit to anything.