# See the whole business. Earn the first deployment.

> Aumenza's method for moving from a business-wide AI opportunity map to one controlled production workflow, with evidence and a go, revise, or stop decision at every stage.

Canonical: https://aumenza.ai/method · Last updated: 2026-08-30

Method / From opportunity map to operation

We assess across functions, then deliberately narrow to the first accountable workflow—and let the evidence earn each expansion.

Four stages · four gates

Each stage must end with evidence, an owner decision, and explicit permission to proceed, revise, or stop.

## Think broadly. Make the next decision smaller.

The four-stage loop

The goal is not to force every opportunity into production. The goal is to choose the best first move, test it honestly, and let operating evidence earn any broader transformation.

### Assess and choose the first move

Opportunity Scan

Map how work, decisions, data, and systems move across the business, rank the opportunities, and choose one accountable workflow for a well-scoped deployment.

Exit evidence: A prioritized opportunity portfolio and one first workflow with usable data, an owner, a baseline, and a safe fallback.

### Prove it in shadow

Readiness Pilot

Build the working system, replay historical examples, run it beside the current process, and compare quality, cost, and speed.

Exit evidence: Pre-agreed value, quality, reliability, and safety thresholds are met—or the project stops or changes.

### Deploy with control

Production Deployment

Connect production identity and permissions, add approval and rollback paths, instrument the system, and release access in stages.

Exit evidence: Business, technical, and security owners accept the operating evidence and recovery plan.

### Operate and improve

Managed AI operations

Monitor value, quality, cost, model changes, exceptions, and adoption. Expand only when the first workflow continues to earn it.

Exit evidence: A durable operating rhythm, documented ownership, and a clear next workflow—or a clean handoff.

## Four gates. One accountable decision.

Production readiness

Thresholds are agreed before the result is known.

- GATE 01 **Value.** Does the measured gain justify the cost and change?
- GATE 02 **Quality.** Does it meet the task-specific acceptance threshold?
- GATE 03 **Reliability.** Does it behave under load, exceptions, and failure?
- GATE 04 **Safety.** Are permissions, approval, audit, and recovery appropriate?

## One workflow. Essential integrations confirmed after discovery.

The first engagement

A fixed-scope, paid engagement starts with the current process and its economic baseline. Historical replay and shadow mode come before consequential writes. At least one human approval point is required, with additional controls based on the workflow's risk.

Scope, price, timing, and acceptance thresholds depend on the workflow and are confirmed in a written engagement.

[See the method applied to a workflow](https://aumenza.ai/demonstration#transformation)

[Discuss your workflow](https://aumenza.ai/start)