EO Decision Engine
Replace the spreadsheet, the custom script or the old point tool. Models update with your data, and every answer shows its reasoning.
eode.ai
EO Decision Engine
Ask a planning question in plain English. You get the best plan your constraints allow, and the reasons behind it.
Modeling core
Enterprise Optimizer®
The rules that turn your data and your question into an optimization model, built up over 25 years. It runs under EO Decision Engine and VCO.
AI assistant
RIA
Explains models and results and suggests changes. Nothing changes until you accept it.
Value Chain Optimization
VCO
Our value-chain planning application, built with Accenture and customers such as Philip Morris. It runs on Digital Planning Twin™, our Azure platform.
The language model never writes equations. RIA drafts a design from your question and data, you confirm it, and Enterprise Optimizer®'s rules generate the math. Every build is checked against the design you confirmed.
Your question
In plain language
Your data
Excel workbooks today
The design
RIA drafts it; you confirm it
What we build
Enterprise Optimizer®
Its rules turn the confirmed design into the optimization model
Gurobi
Solves the model and certifies the optimum
The decision
Trade-offs and the binding constraint, traced to the inputs
Start from your workbook, or from a model spec your Python script exports. You don't start from scratch.
- 01
The spreadsheet
It breaks every time the rules change. You rebuild formulas, re-validate cells, and the business still isn't sure it can trust the answer.
- 02
The custom optimization code
It was right the day it shipped. Now it's brittle, the person who wrote it has moved on, and any change takes weeks.
- 03
The legacy optimization app
It solved one problem five years ago. The business kept evolving; the application didn't.
Example decision
You ask
We need to cut logistics cost ~12% next quarter, but we can't drop below our 98% on-time service level. What should we change?
Constraints it must hold
- On-time service ≥ 98%Waiting
- Existing carrier contracts onlyWaiting
- No new capexWaiting
Plans it tests
- Move all air freight to intermodalQueued
- Re-route 3 lanes via Memphis, move 14% of air to intermodalQueued
- Logistics cost
- -12.4%
- On-time service
- 98.3%
- CO₂e
- -9%
The 98% service floor is the binding constraint: savings stop where it would be breached.
Your logic, preserved · trust, built in
Your spreadsheet or optimization code holds years of business rules, constraint definitions, objective functions and refined scenarios. The engine reads Python optimization code and the Excel models feeding it, and translates them into a decision intelligence application that's faster, easier to maintain and far more engaging for business users.
What carries over
- Existing constraints and objectives become engine models automatically
- Business rules refined over years are carried forward, not thrown away
- Rules change through conversation instead of code rewrites or formula archaeology
- Nobody has to babysit the original script forever
Every number, traceable to its source
Excel shows you a formula. The engine shows the whole causal chain: what fed a result, which constraint shaped it, and how an upstream change would move it, in plain business language.
- Drill any result down to its data, constraints and rules
- Versioned, governed, audit-ready models
- Role-based views for the CFO, the planner and the analyst
- Personal library promotes into the corporate library under controls
- Food & beverage
- Manufacturing
$5Min profit
nobody had found, across five linked models
Five linked models, from long-range capacity down to daily truck loading, showed that not every plant should build every product near its customers.
- Global manufacturing
- 40+ sites
$500M+identified savings
in six months, across 40+ manufacturing sites
A ten-year planning model covering 40+ manufacturing sites moved planning from annual to monthly and cut scenario turnaround from weeks to days.
- Healthcare
- Hospital operations
24beds closed
and $2.3M CAD saved, with no patient turned away
A model of patient flow across oncology, surgery, medical and the ED showed which beds could close safely, and which popular ideas weren't worth the cost.
References: Ash Grove · CBB Metal · Philip Morris · Georgia-Pacific
- Spreadsheets
- Can iterate toward a plan, but can't show it's the best one.
- Custom code
- Hard to change, and tied to whoever wrote it.
- Point tools
- Built for one slice of the business.
- Generic AI
- Independent OR benchmarks put structurally correct optimization models at roughly 28–55%.*
- EO Decision Engine
- Plans that meet every constraint you state, with an optimality certificate and a trace behind each number.
* ORLM (2024), ORGEval (2025).
In our Azure cloud, or self-hosted in your own AWS account.
Our cloud
Hosted on Microsoft Azure. Each organization's data is kept separate and never used for training.
Your AWS account
Self-hosted in your own AWS account, by invitation.
Your assistant
Bring ChatGPT, Claude, Gemini, Microsoft CoWork/CoPilot, or another provider, under your existing AI governance.
Your data
Excel workbooks today. More connectors are coming.
Workflows · one platform
From one-off answer to repeatable process
- Drag-and-connect workflow builder
- Scheduled and event-driven re-optimization
- Hand-offs and approvals built in
One platform for every planning problem
- Same platform for every planning use case
- One team builds and supports multiple applications
- Shared models and data across decisions
- IT supports one vendor, not five point solutions
Customers and partners
Bring a spreadsheet, a script or a plan that takes weeks to run. We'll work through it with you, live.
