Service 06 · Phase Two · Entry gate
006 / 08
Process · design
6–8 weeks

Find the workflow. Design the new version.

Phase Two begins here. Six to eight weeks to find the workflow worth changing and design exactly what the new version looks like — AI role, human role, review points, and exception handling — before anyone builds anything.

Phase Phase two · entry gate
Engagement Fixed scope · two-stage
§ 01 · Overview
Service Process Analysis and Workflow Design
Phase Phase two · entry gate
Prerequisite Phase One complete or equivalent
Duration Six to eight weeks
Format Process mapping + design workshops
Team Principal consultant
Workflows analysed 2 to 3
Workflows redesigned 1
Output Complete workflow specification

Not every manual, repetitive, or painful task is an AI problem. And knowing that a workflow needs AI is not the same as knowing how to change the work. Both questions need answering before a pilot can begin.

This service runs two stages inside one engagement. Stage One goes deep on two to three candidate workflows — maps every step, scores every pain point, and sorts everything into three categories: what AI can genuinely help with, what should be fixed without AI, and what is not worth solving right now.

Stage Two takes the selected workflow and redesigns it completely: what AI does, what people own, where review happens, what good output looks like, and how exceptions are handled. The result is a complete workflow specification — the design brief for anyone who builds or pilots the tool. This service produces the design. Not the build.

§ 02 · Service map

Two stages.
One specification.

Stage One selects and maps. Stage Two redesigns. The output is a workflow specification complete enough for an internal team or implementation partner to build from.

Inputs · workflow · data
Outputs · decision · handover
Workflows analysed 2 to 3
Workflow redesigned 1
Duration 6–8 weeks
Final artefacts 3
§ 03 · Fit

Right for you if
Phase One is done.

This layout is for explaining a service as a repeatable operating shape: who it is for, how the work moves, and what the client leaves with.

001

Phase One is complete.

The data layer, governance layer, and literacy layer have been assessed. The business knows where its foundations sit. Now the question is where to apply AI first and what the new process should look like.

002

You have candidate workflows in mind.

Leadership has identified two or three processes causing friction. Stage One validates which is the right one to change and whether AI is actually the correct solution — or whether the process just needs to be fixed.

003

You need something to build from.

An internal team or implementation partner will run the pilot. They need a complete, unambiguous specification to work from — not a concept or a slide deck.

§ 04 · How it works

Six to eight weeks.
Two stages.

Stage 01 · Weeks 1–2

Select and map.

Identify two to three candidate workflows with leadership. Define scope and selection criteria, then map each workflow end-to-end: every step, every decision point, every handoff, and everywhere things go wrong.

Weeks 01–02
Stage 01 · Weeks 3–4

Score and sort.

Score every pain point against AI fit, business impact, data availability, and owner readiness. Produce the AI Opportunity Shortlist, Process-Improvement-First List, and Deprioritised List. Select the one workflow to redesign.

Weeks 03–04
Stage 02 · Weeks 5–6

Define the roles.

Design the AI role — what it does, what it produces, its operating limits. Design the human role — what people own, what they review, and what they decide. Map the redesigned workflow with roles clearly separated.

Weeks 05–06
Stage 02 · Weeks 7–8

Specify and deliver.

Document every review point, quality standard, exception path, and failure mode. Present the complete workflow specification to leadership and confirm it is sufficient to brief whoever builds or runs the pilot.

Weeks 07–08
§ 05 · Outputs

What the client
leaves with.

Process Maps

A step-by-step map of each analysed workflow: who does it, what information moves through it, and where it breaks down.

Proof

Makes the real workflow visible before any redesign begins.

AI Opportunity Shortlist

The specific pain points where AI fits, ranked by usefulness and readiness.

Proof

Identifies where AI can genuinely help.

Process-Improvement-First List

The pain points that should be fixed with a better process, template, checklist, or system — not AI. Often the most valuable output.

Proof

Prevents tool-first waste by identifying what AI should not touch.

Deprioritised List

What is not worth solving now and why.

Proof

Makes the no-list explicit and defensible.

Redesigned Workflow Map

The new process for the selected workflow, with AI steps and human steps clearly separated.

Proof

Shows exactly how the work changes before anything is built.

AI Role and Human Role Descriptions

What AI does in the new workflow and what people own, review, decide, and escalate.

Proof

Keeps judgement and accountability explicit from day one.

Quality Standards and Acceptance Criteria

What good output from this workflow looks like in plain English, and the specific tests that would confirm the redesigned workflow is working.

Proof

Gives staff, builders, and leadership the same standard.

Pilot-Readiness Recommendation

A clear view of whether the workflow is ready to pilot and what would need to be true before a build begins.

Proof

Creates the bridge to Service 07.