Scope a 6-week pilot
Scope a 6-week pilot

How it works

Start contained. Prove value. Scale what works.

A contained pilot, not a transformation programme.

We begin with one business domain: a corpus, workflow, reporting pack or decision process where scattered knowledge is slowing teams down or blocking AI adoption.

The pilot method

Six weeks, four steps, one success test.

One corpus. One workflow. One success test. Fixed scope, fixed price and a clear handover — with an optional path into a production build if the pilot earns it.

  1. 01

    Scope

    Week 1
    Focus

    Select the corpus, workflow, sponsor and success test.

    Output

    Pilot charter, source list, evaluation questions.

  2. 02

    Structure

    Weeks 2–3
    Focus

    Model entities, metrics, definitions and relationships.

    Output

    Entity map, metric map, relationship model.

  3. 03

    Build

    Weeks 4–5
    Focus

    Build the graph prototype and retrieval workflow.

    Output

    Source-linked answer interface.

  4. 04

    Evaluate

    Week 6
    Focus

    Test answers and assess value.

    Output

    Quality results, gaps, go / no-go decision.

  5. +

    Extend

    Optional · Weeks 7–8
    Focus

    Define the production path.

    Output

    Architecture, governance and scale roadmap.

Principles

The rules behind every engagement.

01

Start contained

One corpus, one workflow, one success test. Value is proven before scope grows.

02

Govern before you generate

Definitions, ownership and controls come before AI features.

03

Source-linked or it does not ship

Every accepted answer must trace back to evidence.

04

Build for handover, not dependency

You keep the model, documentation and roadmap — no black box.

Questions

Straight answers before you commit.

How much does a pilot cost?
Pilots are fixed-scope and fixed-price. The price depends on corpus size, workflow complexity, security requirements and the level of prototype required. The purpose is to prove value before committing to a larger build.
What do you need from us to start?
A bounded corpus or workflow, one accountable sponsor, access to the relevant documents or data sources, and 30–50 questions or decisions the system should support.
Is our data safe?
The pilot is designed around your environment, approved access methods and agreed security boundaries. Sensitive production data is not always required for an initial pilot.
What if the pilot shows it is not worth continuing?
Then the pilot has done its job. You keep the findings, semantic model, evaluation results and roadmap. The goal is to test value before scale.
Do we become dependent on Amidflow?
No. Every pilot includes clear documentation, model definitions, handover notes and an extend-or-stop recommendation. We avoid black-box dependency by design.

Start

Bring one corpus, workflow or reporting problem worth scoping.

One bounded domain where scattered knowledge is slowing decisions down or blocking AI adoption — and where a source-linked answer would create measurable value.

Scope a 6-week pilot

One scoping call. No deck, no prep.