eode docs
Getting started

Core concepts

The vocabulary eode actually uses: models, datasets, Base, scenarios, workspaces, workflows, and RIA.

RIA

RIA is eode's assistant. It runs the clarifying interview, authors and edits models, turns chat messages into what-ifs, proposes workspaces and workflows, and reviews results. The RIA Console is the panel down the right of the screen — and one rule governs it everywhere: RIA proposes, you approve. A change arrives as a Plan confirm card with a Run plan button and a diff you can open; where the change supports it, Edit and Cancel sit beside it. Nothing is applied silently.

Model

A model is eode's working representation of your decision: products, resources, processes, demand, capacities, costs, and the objective you are optimizing. You build one in five steps — describe, clarify, build, load data, solve — either Basic, where you describe the business in words and RIA authors it with you, or Advanced, where you bring a model or a data pack you already have.

Models live in the Model Library in two tiers: Personal (yours) and Organizational (shared with your organization, with versions, locking, permissions, and an audit log). Every model has a home page — where it stands, what it's worth, and every door into it.

Model Design

The structure RIA proposes before anything is solved: what's in scope, the goal, and the rules and choices in plain English. You check it on one page — Confirm or Needs correcting per section, then Accept the design — or send it back with notes. The accepted design stays on the model's home page as a record you can download, and Send back reopens the questions with your answers kept.

Dataset

Data you have imported, listed under Datasets: "Import once, preview it any time." A dataset is made of resources — a sheet, a table, a view — each with its own rows, previewable and editable in a grid. One dataset can be attached to any number of models; a model consumes either your own dataset or sample data.

Base and scenarios

A model's accepted data and its original solve are the Base scenario — read-only, the reference point. A scenario is a what-if branched from it: "What if flour costs jump 30%?" Every scenario carries its own solve, its own KPI values, and deltas against Base. Scenarios are cheap — on the Scenarios screen a chat message becomes one — and disposable, comparable, and verifiable. A model imported from a workbook also starts with BASE DATA: the imported values as stated, before any solve.

KPI

KPIs are the numbers that matter for your decision. RIA proposes them from the model's objective and constraints; you accept, add, or remove. Each tile carries its own confidence — VERIFIED or ESTIMATED — and traces to where the number came from. See KPIs & custom metrics.

Workspace

A saveable analysis canvas of widgets — metrics, charts, tables, maps, narratives — bound to one or more scenarios. Build one by hand, from a template, or by describing what you want to RIA. See Workspaces.

Workflow, run, stage, gate

A workflow is a published recipe: stages with widgets, roles, and a gate between stages — a stage either advances manually or waits for a sign-off. A run is one execution of it: assignees fill stages, approvers sign the gates, and scenarios and analyses attach as evidence. See Workflows & runs.

Revision and change order

Every structural change to a model lands as a numbered revision, with what it did to the objective and a way back. A change order is a proposed change sitting on the model's Develop backlog in business language — previewed before it is applied, and set aside only with a recorded reason. See Changing a model.

Validation

One standing record per model, in three pillars — Data, Functionality, Results — each with its own verdict: Validated, Warnings, Not passing, Stale, or Not validated. A verdict that no longer describes the current model is marked Stale rather than served as current. Per-scenario evidence is a separate thing: that's Verify. See Validation & trust.

Personas and roles

Your persona — Modeler, Analyst, Operator, or Executive — shapes the product around how you work: the Home greeting, what Home leads with, whether you are offered authoring, and how RIA opens a conversation. It does not gate any screen.

Two separate sets of roles do govern what you can do. Platform roles — Member, Promoter, Org Admin — decide who can promote a model and who administers the platform surfaces. Organization roles — owner, admin, member — decide who runs the organization itself: inviting and removing members, renaming, ownership. See Accounts & access.

On this page