# EcoKure: one-page enterprise buyer brief

## The simple proposition

EcoKure is a clear control layer around the AI a company already uses. It
helps an accountable team set the rules, send uncertain cases to a named human
reviewer, and preserve a record that can be checked later.

## The product hierarchy

`AI Control Tower` → `Governed Workflow` → `Assurance Runtime` → `Control Pack` → `EvidencePack`

- **AI Control Tower** — the product people use to see decisions, review and change impact.
- **Governed Workflow** — the customer workflow being evaluated.
- **Assurance Runtime** — the layer deployed in the agreed customer boundary.
- **Control Pack** — the customer-owned rules, evidence requirements and reviewer roles.
- **EvidencePack** — the signed, replayable record of what happened and why.

## The first worked pilot: life-sciences evidence change control

When a scientific source, dataset or protocol changes, a team needs to know what
the change affects without redoing everything or silently relying on old
evidence.

```text
Source update → dependency impact → Control Pack check → human review
             → selective replay → signed before/after EvidencePack
```

The pilot is non-diagnostic and runs in shadow mode first. It demonstrates one
named workflow, one owner, one controlled change, a review queue, a signed pack
and a written go/no-go decision.

## What success looks like

1. The workflow boundary and owner are agreed.
2. The Control Pack is versioned and reviewable.
3. A source change produces an affected / preserved / unknown impact set.
4. Unknown cases reach a qualified reviewer with a reason and disposition.
5. The affected set is selectively replayed.
6. The EvidencePack verifies independently and records limitations.

## What EcoKure does not do

EcoKure does not diagnose patients, approve credit, move money, replace model
validation, replace a safety system, determine legal compliance or make the
customer's accountable decision. It provides control and evidence around a
bounded workflow.

## Current proof and its boundary

The AWS target-representative run recorded 117,290 completed verifications, zero
performance errors, KMS-backed signing, PostgreSQL persistence, controlled
recovery with no lost acknowledgements and clean Terraform teardown. This is
target-environment evidence, not customer production validation, certification
or an AWS endorsement.

## The review decision

The useful first conversation is not “can EcoKure govern everything?” It is:

> Which one workflow has a clear owner, changing evidence and a costly review
> or audit problem—and what would we need to see before expanding it?

The next step is an Evidence & Control Readiness Evaluation around that one
workflow.