Delivery route optimization
Kohepets gives a delivery agent the tools to investigate routes and plan around disruptions.
A delivery agent uses mapping, traffic and event tools to assess routes, anticipate recurring disruptions and compare alternatives. It plans around delivery commitments and revisits its recommendations when conditions change.
The impact
24%
less distance per delivery route
Delivery route optimization
Case study facts
- Industry
- Pet retail & pharmacy
- Team
- Operations
- Systems connected
- 4
- Time to live
- 2 weeks
- Engagement
- AI implementation
The impact
Results for the team
- less distance per delivery route
- 24%
- less distance per delivery route
- Route recommendations reduce delivery distance while accounting for commitments, traffic, restrictions and recurring events.
- less dispatch planning time
- 78%
- less dispatch planning time
- The agent investigates routes and checks alternatives, reducing the time dispatch spends preparing a plan.
- deliveries within the planned window
- 96%
- deliveries within the planned window
- The delivery rate tracks how many deliveries arrive within their planned window, with affected stops reassessed when conditions change.
Situation
The same roads can need a different plan each day.
Delivery planning depends on more than addresses and travel times. Recurring events, road restrictions and changing traffic can make an otherwise sensible route unreliable. Dispatch needs to check what affects the day's deliveries and decide which alternatives are actually workable.
Challenge
A route needs context, not just a shorter distance.
The relevant checks vary by route and delivery time. A recurring event may affect only certain streets or hours, while a diversion can create a conflict with a customer's delivery window. The agent needs to decide what to investigate, assess the evidence and test alternatives, while respecting vehicle capacity and existing commitments.
What we built
An agent that investigates, compares and replans.
We gave a delivery agent tools to inspect orders, query maps and traffic, check event information and evaluate route alternatives. It identifies where a plan needs more investigation, chooses the relevant tools and assesses whether recurring events or restrictions overlap with the proposed journey.
It then uses a routing solver to compare feasible alternatives, revisiting its plan when a check reveals a conflict. Dispatch receives a recommendation with the reasons for route or timing changes and any unresolved uncertainty.
- 01
Understand the delivery goal
Review destinations, delivery windows, vehicle capacity and driver availability to establish what a workable plan must satisfy.
- 02
Choose what to investigate
- Tool selection
Identify routes, locations and time periods that need traffic, restriction or event checks, then select the relevant tools.
- 03
Assess events and conditions
Check whether recurring events, closures or congestion are relevant to the route and scheduled delivery time.
- 04
Test alternative plans
- Tool calling
- Constraint solver
Use mapping and route-solving tools to compare different sequences, departure times and diversions against delivery constraints.
- 05
Resolve conflicts and explain
- Self-correction
Run further checks or revise the proposal when an alternative creates a new conflict; surface uncertainty and explain the recommendation.
- 06
Reassess as conditions change
Investigate new disruptions or delivery updates and propose targeted changes to the remaining route for dispatch review.
Evals and guardrails
Verify tool use. Check every route.
Tool-use evals
Test whether the agent chooses the right tools and supplies correct inputs. Include outdated traffic, recurring events and tool failures.
Route evals
Use the routing solver to check proposed plans. Measure how often routes meet vehicle capacity, delivery windows and road restrictions.
Guardrails
Check source dates and tool results, limit retries and block routes that break constraints. Require dispatch approval before release.
Results
Dispatch gets routes backed by current context.
Route recommendations reduce delivery distance while accounting for commitments, traffic, restrictions and recurring events. The agent investigates routes and checks alternatives, reducing the time dispatch spends preparing a plan.
The delivery rate tracks how many deliveries arrive within their planned window, with affected stops reassessed when conditions change.
- less distance per delivery route
- 24%less distance per delivery route
- less dispatch planning time
- 78%less dispatch planning time
- deliveries within the planned window
- 96%deliveries within the planned window
From Kohepets
“We have worked with Revensi for many years. In that time they have made vast improvements to how our business runs, and they are great people to work with. We highly recommend them.”
Build Capacity To Grow. Own Your Intelligence.
Bring the workflow slowing your sales, orders or delivery. We'll map the bottleneck and show you what the AI system could look like.