Case Study

A Leading Vietnamese Beverage Company Moves 22% More Orders Per Trip, Slashes Fuel Consumption by 37%

The Locus agentic TMS runs the company's distributor delivery as one decision layer: route planning, stop sequencing, mid-route re-optimization, and trip-close reconciliation, all against the distributor's own rules.

Download Now

Segment & Geography

Industry: CPG, Beverage
Region: Vietnam

Objectives

Automate the daily route planning and trip-close reconciliation that ran manually.
Give distributor owners a live view of every route and delivery in progress.
Increase orders per trip while reducing distance and fuel consumption.

What Locus Runs

Automated route planning and stop sequencing across the distributor's mixed fleet.
Address validation for the loose addresses small retail points carry.
Mid-route re-optimization on live data, with the revised sequence on the app inside the trip.
Proof of delivery and payment capture at each stop, on the delivery staff's phone.
Automated reconciliation of money, goods, and delivery results at trip close.
A live dashboard of vehicle position, completed stops, and open stops.

Impact

↓37%

Fuel consumption

↑22%

Orders per delivery trip

↓60%

End-of-day reconciliation time

↓35%

Route planning time

Client Overview

One of Vietnam's largest beverage companies reaches the market through a network of distributors. Each distributor runs a single depot out to thousands of small retail points a day, on a mixed fleet of vans, trucks, and motorbikes. Every shop keeps its own opening hours, access constraints, and order frequency.

For years, distributor delivery relied on manual planning and reconciliation, and the cost was borne by the distributors. Distributors raised it as a constraint on their own growth.

After seeing the operation for itself, the company brought in Locus to automate the delivery decisions its distributors had been making manually. Locus is the world's first agentic TMS, automating delivery and logistics decisions since 2015.

Business Challenges

Route planning ran manually in a spreadsheet. Each morning a planner built the day's routes in Excel, working round by round and salesperson by salesperson across hundreds of orders. It took an hour to an hour and a half before any vehicle moved. Any change to an order or a vehicle meant rebuilding it.

Retail point locations were not held in any system. Many small shops had no validated delivery location, so delivery staff located them en route, calling sales and marketing for directions. Time that should have gone to serving the stop went to finding it.

No single view of the fleet in motion. With vans, trucks, and motorbikes on the routes at once, the distributor had no live picture of vehicle position or delivery status. Building one meant asking each staff member for an itinerary.

Trip-close reconciliation ran manually. Money, goods, and delivery results were matched by staff at the end of each route. The work was error-prone, and finalizing the day's data often carried into the next day.

Implementation Approach

On-site before anything was configured. Forward-Deployed Engineers worked on-site for weeks with each distributor, covering route planning, salesperson territories, delivery addresses, and vehicle utilization. Each distributor's business rules and end-of-day reconciliation were documented from the operation itself.

FMCG requirements became configurable planning rules. This went past a standard route optimization deployment. Partial deliveries, promotion-driven order fulfillment, vehicle restrictions, and multi-drop routing were modeled as planning rules. Customer-specific delivery constraints and distributor-specific planning logic were configured per distributor, matching the way each already worked.

Planners validated the configuration before deployment. Planners worked through the solution configuration ahead of go-live, checking the optimization against real operational scenarios. Anything that missed how the depot actually ran was changed before the first live route.

Side by side through go-live and hypercare. During go-live, Locus worked with planners and drivers on hands-on training, route validation, and issue resolution, refining the configuration on operational feedback. The team stayed engaged through hypercare, monitoring adoption and resolving issues as each distributor came online.

Planners and delivery staff were running the system themselves by the end of each rollout.

How It Works

The Locus agentic TMS runs as the distributor's system of execution. Orders stay in the company's DMS as the system of record. Locus takes each day's orders from there and turns the delivery decisions into a plan the team runs. The company and its distributors set the rules.

Route planning runs automatically against the distributor's rules. Locus Agents take the day's orders from the DMS and plan every route against hundreds of live constraints. Those constraints include the time windows, vehicle types, and access rules each retail point carries. They return a sequenced plan with pick lists and trip assignments. An added order or a changed vehicle re-plans the day without a manual rebuild.

Every retail point resolves to a validated delivery location. Address validation pins each shop from the loose addresses these retail points carry. Stops arrive in order on the delivery staff's phone with the order detail for each. When conditions change mid-route, the system re-sequences on live data and pushes the revised order to the app.

One live view of every vehicle and every stop. A dashboard shows vehicle position, completed stops, and open stops as the day runs. The distributor reassigns delivery staff between trips and resequences stops from that view while the day is still running.

Reconciliation completes at trip close. Proof of delivery and payment are captured at each stop on the app. Money, goods, and delivery results are matched into one traceable record as delivery staff return, and the day's data is finalized the same day.

Every autonomous decision stays governed by six mechanisms: explainability, traceability, evaluation, autonomy levels, execution sandbox, and human-in-the-loop override.

The Results

People

Planning and reconciliation moved into the platform.

35% less time on route planning.
60% less time on trip-close reconciliation.
Delivery staff run a sequenced route from the app, with no calls to locate a shop.