Decision Optimization Platform.

Production and material readiness decisions

Which production plan should the company authorize when material, labor, sequence, and due-date constraints do not line up?

Kimaru brings the evidence, judgment, constraints, and AI recommendations for one consequential decision into a governed loop so teams can compare feasible choices, put the commitment with the right authority, execute the approved action, measure the result, and preserve what the company learns.

Start with one consequential decision.

Which recurring decision is costing your operation the most?

Systems, people, and AI contribute to one Kimaru Decision Model. The company compares choices, approves an action, measures the result, and keeps governed learning for the next comparable decision.

One recurring decision, kept together from current reality to measured outcome.

Decision Optimization for the decisions that matter.

Your systems provide facts. Your people provide judgment. Your models and AI provide intelligence. But none of them owns the consequential decision end to end.

Kimaru brings together the facts, judgment, rules, and intelligence a decision needs, models the available choices, and optimizes across the objectives and constraints that matter to you.

You can compare feasible alternatives and their expected consequences before committing to an action. Kimaru then carries the approved decision into execution, measures the outcome, and preserves what was learned.

Model the choices. Optimize the trade-offs. Govern the commitment. Measure the outcome.

Keep your ERP, planning systems, models, and AI. Kimaru keeps the decision together.

Facts Judgment Rules AI KIMARU Optimized decision Feasible choices compared against real objectives GOVERNED COMMITMENT

Measured customer proof

Live decisions. Measured outcomes.

A live production-planning Decision Model at an aircraft-carpet manufacturer, measured against operating performance and ERP validation.

43% to 75%

New-order on-time production

96% to 100%

ERP-versus-plan pass rate, WK24 to WK26

When the normal plan stops working

The hardest decisions do not live in one system.

Enterprise systems record orders, inventory, materials, capacity, shipments, and other operating facts. But the exceptions, trade-offs, current context, and practical rules that change a decision often sit somewhere else, or with people.

When a consequential decision crosses those boundaries, another recommendation is not enough. You need to bring the relevant facts and judgment together, compare and optimize feasible choices against the outcomes that matter, make the commitment with the right authority, act, and measure what happened.

Enterprise-system facts

Orders, inventory, materials, capacity, costs, plans, and operational state.

Employee judgment

Current context, exceptions, experience, corrections, and reasons the system record is incomplete.

Business rules and constraints

Policies, thresholds, priorities, commitments, approval rights, and non-negotiable boundaries.

AI recommendations

Interpretation, prediction, simulation, optimization, and proposed actions.

Decision Model

A reusable Decision Model defines how one recurring decision should be framed, compared, governed, acted on, and measured.

Decision Object

Each live Decision Object keeps the evidence, feasible choices, authority, commitment, action, outcome, and applicable learning together for one actual decision.

Decision Receipt

The Decision Receipt records what was known, what was assumed, which alternatives were considered, what was selected, who authorized it, and what result was expected.

Why now

Turn more intelligence into better operating decisions.

AI is making intelligence abundant and giving software more ability to act. But more intelligence does not automatically produce better business outcomes.

The value comes from turning distributed intelligence into good decisions, carrying those decisions into action, measuring what happened, and learning from the result.

As intelligence becomes abundant, the advantage shifts toward Decision Capacity: how well you convert intelligence into consequential decisions and measurable outcomes.

Kimaru is built to increase it.

As AI agents take on more work, this becomes more urgent. ERP agents, planning agents, pricing agents, employees, and executives may all contribute different recommendations and objectives. Kimaru gives you a common decision structure without requiring any one system or agent to own the decision.

Abundant Intelligence AI, agents, models Decision Capacity KIMARU Measured Outcomes Value, learning

One recurring decision, kept together

From uncertainty to measured outcome.

1

Understand current reality

Bring together current operating facts, missing context, constraints, assumptions, and sources of uncertainty.

Understand current reality
Kimaru Work Hub showing recurring operational decisions and recommendations ready for review by the responsible team.
2

Model what drives the outcome

Make controllable actions, outside conditions, intermediate effects, and business outcomes visible in one decision structure.

Model what drives the outcome
Kimaru Causal AI Modeler connecting controllable actions, external conditions, intermediate effects, and outcomes for a truck-load planning decision.
3

Compare feasible futures

Keep realistic choices and their expected consequences visible long enough to examine the tradeoffs before commitment.

Compare feasible futures
Kimaru Decision Review showing recommendations for a truck plan and the operational evidence available before selection.
4

Govern the commitment

Place the decision with the named owner. Capture approval, adjustment, deferral, rejection, or escalation and preserve the commitment in a Decision Receipt.

Govern the commitment
Kimaru approval view showing accepted, adjusted, and deferred recommendations with controlled options for passing an approved result to existing systems.
5

Act within agreed authority

Pass the approved action or executable plan to people, agents, APIs, or existing systems only within the agreed integration and approval boundary.

Act within agreed authority
ERP, planning software, spreadsheets, suppliers, people, and AI agents contribute to one recurring decision that Kimaru keeps together as an outcome-linked record.
6

Measure value and learn

Compare the expected and actual result. Establish Decision Value when the economic change can be attributed, then preserve, revise, or retire the learning for the next comparable decision.

Measure value and learn
Kimaru Decision Tracker showing operational outcomes, comparison baselines, and decision history for truck-load planning.

Start with a decision someone already owns

Which consequential decision should your company improve first?

Production and material readiness decisions

Which feasible production and customer-commitment plan should the company authorize when material, equipment, labor, sequence, and due-date constraints do not line up?

Examples of choices to compare
Build, resequence, split, substitute, transfer, expedite, defer, or escalate.

Outcome measures to define with the customer
On-time production, throughput, schedule stability, expedite cost, service, or revenue protected.

Supplier, BOM, and cost decisions

Should the company buy, wait, expedite, substitute, transfer, renegotiate, or escalate when readiness, quality, timing, cost, and margin conflict?

Examples of choices to compare
Buy, wait, expedite, substitute, transfer, renegotiate, qualify an alternative, or escalate.

Outcome measures to define with the customer
Purchase-price variance, margin, material availability, continuity, quality, lead time, or avoided expedite cost.

Inventory and allocation decisions

Where should constrained inventory go, and which revenue, service, working-capital, and operating-risk tradeoff should the company accept?

Examples of choices to compare
Allocate, hold, transfer, replenish, reserve, substitute, rebalance, or escalate.

Outcome measures to define with the customer
Revenue served, service level, working capital, stockout exposure, obsolescence, or operating continuity.

Kimaru is not limited to these three decisions. They are concrete starting points on one Enterprise Decision Operating System. Once the first Decision Model proves value, the next connected decision can reuse relevant evidence, rules, authority, and governed Decision Memory.

Measured customer proof

One production-planning decision created measurable value.

An Aircraft Carpet Manufacturer used Kimaru for a recurring production-planning decision. ERP and spreadsheet data had to be reconciled with material readiness, loom compatibility, lot-mixing rules, capacity, due dates, and planner judgment.

Kimaru did not replace the manufacturer’s ERP or planners. A reusable Decision Model made the constraints and feasible choices inspectable. Planner corrections and their reasons remained part of the decision. The approved plan could be validated against ERP records, and the Decision Tracker kept expected and actual results linked to the decision that produced them.

Aircraft Carpet Manufacturer

The measured results are summarized above. Results are not universal. The ERP-versus-plan rate is a validation metric, not an on-time production or delivery rate.

A Kimaru Decision Model contains the uncertainty, evidence, human context, rules, causal map, feasible choices, expected consequences, authority, decision receipt, actual outcome, Decision Value, and governed memory.

The Decision Model keeps uncertainty, evidence, constraints, authority, expected result, actual outcome, and learning connected.

Keep the enterprise stack

Keep the enterprise stack. Add the decision layer.

ERP remains the system of record. Planning and optimization systems keep doing what they do best. AI and agents contribute intelligence and action.

Kimaru connects those capabilities around the consequential decision.

AI models

AI models

Interpret evidence, predict, simulate, optimize, and propose. They contribute intelligence but do not inherit enterprise authority unless that authority is explicitly delegated and governed.

Agents

Agents

Retrieve, coordinate, recommend, or execute bounded work within the access and authority they are given.

Enterprise systems

Enterprise systems

Preserve transactions, plans, records, operational state, and authorized execution in their domains.

Kimaru

Kimaru

Keeps the complete recurring decision together across those contributors, including evidence, alternatives, human judgment, authority, commitment, action, outcome, value, and governed learning.

Examples of systems, models, and agents that can participate when configured and authorized.

SAP
S/4HANA
Oracle
Microsoft
SAP

Joule

Copilot
Codex
Claude

Keep your existing stack. Kimaru adds the governed decision layer that connects system evidence, human judgment, business rules, AI agents, authorized action, and measured outcomes.

ERP, planning software, spreadsheets, suppliers, people, and AI agents contribute to one recurring decision that Kimaru keeps together as an outcome-linked record.

Kimaru earns its place when the recurring decision crosses tools and organizational boundaries.

Expand from here

Start with one decision. Expand what you can optimize.

Kimaru starts with one consequential recurring decision where better choices have measurable economic value.

Once that Decision Model proves value, you can add connected decisions. Relevant context, outcomes, and learning can carry forward instead of being lost after each decision.

Each completed decision should make the company better at the next one.

+ Decision + Decision + Decision + Decision One Decision LIVE

How it deploys

Start bounded. Prove value. Expand.

  • Start with one bounded consequential decision.
  • Keep the existing enterprise stack.
  • Prove measurable economic value.
  • Expand from there.

The boundary fits the decision

Begin with people in control and the systems of record intact.

ERP and other enterprise systems remain systems of record for their transactions. The named decision owner can approve, adjust, defer, reject, or escalate. Any write-back or external action follows the agreed integration, customer authorization, and approval rules.

A first engagement can begin with controlled files, read-only database views, or scheduled API pulls. It does not require write-back into customer systems.

Kimaru can be configured for customer-managed AWS, a private Kimaru-managed tenant, or an agreed on-premises or private-cloud design. Actual data location, controls, retention, external-model use, and responsibilities depend on the deployment that is contracted and implemented.

Company data, employee know-how, business rules, and the Kimaru Decision Model remain within an agreed deployment boundary while an external AI model is used only for agreed tasks under the selected controls.

External models may assist. They do not become the system of record for the decision.

Start small enough to prove it

Prove one consequential decision first.

Choose a recurring decision where delay or error has a measurable consequence, evidence is split across tools or people, human judgment changes the answer, a named person owns the commitment, and the outcome can be observed in a useful time window.

1

Define the Decision Model and baseline

Name the recurring decision, owner, consequence, evidence, constraints, current process, and value measure.

2

Replay history or run in shadow mode

Produce inspectable alternatives and a governed recommendation without uncontrolled write-back. Compare what Kimaru would recommend with the incumbent path.

3

Capture the commitment and measure the result

Preserve the decision in a Decision Receipt, observe the outcome, establish attributable value where the evidence supports it, and decide whether the result justifies expansion.

  • Measurable consequence
  • Split evidence
  • Human judgment changes the answer
  • Named decision owner
  • Observable outcome

Start with the decision

Which consequential decision should your company improve first?

Start with one recurring decision where cost, revenue, service, or continuity depends on getting it right.

Prefer email? Tell us the decision.