kint horizon
New AI on top. New math underneath. The modern optimization logic under your planning module, producing plans the planners stop overriding.
The problem
Smarter interface, same 2010 math.
Your new AI agents decide faster, but on the same old planning logic. Faster wrong is not better. Planners keep overriding, and the consequences show up in quarterly results.
The optimization logic inside the planning module ignores 2026 reality: tariffs, CSRD emissions as a hard cost constraint, serialization, recall spikes. kint is the new math underneath.
The anchor demo
Re-optimize an existing plan against a new constraint.
Add a tariff or an emission cap. kint re-solves the plan with the real constraint binding, then shows the override-rate drop and the cost impact. This is the planning cockpit.
Planning Cockpit · EU Region
solver: kint engine · gap 0.4%
| Order | Customer | Pri | Req | Alloc | OTIF |
|---|---|---|---|---|---|
| SO-44812 | Nordic Retail | P1 | 3,200 | 3,200 | full |
| SO-44830 | Bavaria GmbH | P1 | 1,800 | 1,800 | full |
| SO-44851 | Iberia Dist. | P2 | 2,400 | 1,650 | partial |
| SO-44877 | Adriatic Co. | P2 | 1,200 | 900 | partial |
| SO-44903 | Baltic Spot | P3 | 2,000 | 0 | deferred |
Planners override less because the model now captures the real constraints. −25 pts override rate.
| target stock · bar = on-hand projection
The modules
Four planning engines. One math underneath.
Each module replaces the optimization logic in a planning function. Same engine, same audit trail, docked under your existing platform.
Release in the right order
Order release and machine sequencing that respect real shop-floor constraints, not heuristics.
Hold less, miss nothing
Multi-echelon inventory optimization (MEIO). Reorder points and target stock solved across the network at once, not per node.
Promise dates that hold
ATP/CTP and allocation under shortage. Promise dates the model can actually keep, with priority logic that holds.
New constraint, no vendor SOW
Set a new constraint without a vendor SOW. Tariffs, CSRD emissions, serialization, recall handling. Switch it on and re-solve.
Your agents decide faster. On the same 2010 math. kint is the new math underneath, so the plan is right, not just quick.
How kint decides
The recommendation, the why, and the delta.
Every plan ships with the binding constraint, its shadow price, and what changed against the old plan. Plus the moves a human would have missed.
Re-optimized plan · EU Region W24–W35
Why horizon
Three things that are true today.
The engine is from 2010.
Your ERP has new AI on top. The planning engine underneath has not changed. We replace the math, not the interface.
Plans that match 2026.
Tariffs, emissions, and serialization as active constraints in the model, not as notes a planner reads and overrides by hand.
The override rate falls.
Planners stop overriding because the model captures the real constraints. The plan is one they can defend.
The angle
Agents on old engines vs the correct math underneath.
The big planning suites all put agents on the same old optimization engines and heuristics. The agent decides faster, not better. kint is the provably correct math the agent calls.
| Their approach | kint horizon | |
|---|---|---|
Enterprise planning suites agentic add-ons | Agents act faster on the existing optimizer and heuristics. | The agent calls provably correct math. Faster and right, not just faster. |
Concurrent planning platforms agent layer | Agent layer over a concurrent planning engine. | kint docks under any platform as the optimization engine. No rip and replace. |
Digital-twin planning diagnostic analytics | Diagnoses value leakage after the fact, in the rear-view mirror. | kint prevents the leakage up front, in the plan itself, with the constraint binding. |
Powered by the kint engine
Provably correct. Defensible. Compliant.
Replace the math, not the interface.
Bring one planning run and the constraint that is hurting you. We will re-optimize it and show the override-rate drop and the cost impact.