talk to usTime windows, driver hours, vehicle capacities and stock levels in one model. kint returns the lowest-cost plan the solver finds under all of them, with its gap, and names the rule that binds.
Tomorrow's deliveries are on the board: trucks, stops, time windows, driver hours, vehicle capacities. By noon the dispatchers have a plan that looks reasonable. Whether a cheaper plan exists under the same rules, nobody can say. And a cancelled stop at four o'clock starts the work again.
Each added stop or time window multiplies the possible routes. Far too many exist to check by hand or in a spreadsheet.
A cancellation or a priority order means redoing the plan by hand. kint re-solves from the current state.
Without a bound, nobody can say how far today's plan is from the best one.
The rules the most experienced dispatchers apply get written down, reviewed and shared with the team.
Each decision comes back as a plan, the rule that binds it, and the gap to the best possible plan. A person releases it before it is used.
Which vehicle serves which stops, in which order, within every time window and driver rule?
Tours per vehicle with stop order and arrival times, the binding constraint, and bound and gap.
A change at one stop is re-solved across the whole plan, not patched locally.
How much stock of each item belongs in which warehouse?
Target stock per item and location, with the service level each one reaches.
Safety stock follows each item's demand variability instead of one fixed rule.
When do we order, and how much, as demand shifts with the season?
An ordering schedule with the lowest cost under supplier limits and the forecast.
Seasonal peaks are planned in the same model as the rest of the year, with their cost stated.
The people stay the same. They work with a plan that states its own quality and its own reasons.
| Dimension | Today | With kint |
|---|---|---|
| Planning | Hours of manual work a day | The engine solves, dispatchers review and release |
| Quality of the plan | Unknown, good enough | Bound and gap, stored with every run |
| Disruptions | Re-plan by hand | Re-solve from the current state |
| Driver rules | Checked afterwards | Hard constraints in the model |
| Record | A spreadsheet on someone's drive | Data, model and release stored per plan |
Exports from your TMS, WMS or ERP are enough to start. kint's decision agent drafts the model from the planners' description. In the pilot, kint's forward deployed engineers (FDEs), who work directly with the customer's team, review it with the experts.
Exports from the TMS, ERP or a database. Every upload is versioned.
Dispatchers state time windows, driver rules and priorities in their own words. The agent writes them down for review.
The plan next to today's plan, side by side, with the gap stated.
A person releases the plan. It goes back only through an API you control; your TMS decides whether to apply it.
A conversation with kint's FDEs. Bring the decision and its rules, and get a first model planners can read.