The Blueprint AAA Framework Who Is This For About FAQ
Document   GEM Blueprint v1.0
Author   Timothy J. Hitchens
Status   Public / Free to use
Backed by   Techshin Partners

// Everyone is talking about AI. Very few are doing it properly.

The GEM
Blueprint.

A free public framework for adopting AI inside your organisation. Not the tools. Not the hype. The governance, the enablement, and the measurement that makes AI actually work.

G
Governance
Who is responsible. What tools are approved. What are the rules. What is the risk. Before anyone touches anything.
E
Enablement
Who is enabled to use AI. Are they trained on approved tools. Do they understand the governance boundaries.
M
Measurement
How you know it is working. What it is costing. What value it is creating. What thin work it has eliminated.
// Section 01 — Overview

The work that needs to happen before anyone starts buying tools or building agents.

Most organisations jump straight to the tools. They install ChatGPT. Someone builds an agent. A team starts experimenting. And nobody asks the three questions that matter: who is governing this, who is enabled to use it, and how will we know if it is working?

The GEM Blueprint is a structured framework for answering those questions first. Then mapping your processes. Then approving the right tools. Then designing workflows. Then deciding what to automate, what to augment, and what is ready for agentic systems.

It is public, ungated, and designed to be shared. A useful starting point for boards, executives, transformation leaders, and operators who need a clear way into AI without pretending implementation is the first step.

It is free. It is practical. And it was built by someone who watched exactly the same thing happen with cloud and spent nearly seven years at AWS watching organisations learn it the hard way.

// Section 02 — What you get

A high-level framework you can actually use.

The GEM Blueprint is meant to be read quickly, shared easily, and used to bring shape to AI conversations before they turn into expensive confusion.

For leaders
A clear executive lens
Use GEM to explain to your board, leadership team, or organisation what responsible AI adoption should look like before implementation starts.
For teams
A shared operating model
Use it to align governance, enablement, workflow design, and measurement so AI does not become another disconnected initiative.
For action
A starting point, not a paywall
Read it, share it, and apply it. If you want help operationalising it inside your organisation, that is where Techshin Partners comes in.
// Section 03 — The Blueprint

Six phases. In order. Do not skip any.

Each phase builds on the one before it. Most organisations start at phase five and wonder why it does not work.

Phase 01
Governance

Before anyone touches an AI tool, answer: who is responsible when something goes wrong? What are the rules? What regulations apply? What tools are approved for use? Governance is the gate. Nothing gets through without it.

Define ownership and accountability for AI decisions
Evaluate and approve individual AI tools before they enter the organisation
Map regulatory, compliance, and insurance requirements
Establish risk framework and escalation paths
Phase 02
Enablement

Who is enabled to use AI? Are they trained on the approved tools? Do they understand the governance boundaries? Enablement without governance is chaos. Governance without enablement is a policy nobody follows.

Identify who needs access and at what level
Build capability through practical training on approved tools
Establish usage policies and permission structures
Create feedback loops so people learn and improve
Phase 03
Measurement

How do you know if AI is working? What is it costing? What value is it creating? What thin work has been eliminated? What valued work has replaced it? If you cannot answer these, you are experimenting, not adopting.

Define what success looks like before you start
Track true cost: tools, compute, time, training
Measure value created, not just activity generated
Build a dashboard your board would actually trust
Phase 04
Process mapping

Before you decide what to automate, you need to know what your people actually do. Most organisations do not have SOPs. They have tribal knowledge and habits. This phase makes that visible.

Document existing workflows and standard operating procedures
Identify thin work: repetitive, low value, machine ready
Identify valued work: creative, strategic, human required
Map decision points within each process
Phase 05
Workflow design

Take the approved tools from Phase 01 and the documented processes from Phase 04. Map tools to workflows. Group them together. Design how the pieces fit. This is architecture, not implementation.

Map approved tools to documented processes
Group tools together to create complete workflows
Identify where human decision points sit within each workflow
Design the workflow before anyone builds it
Phase 06
AAA decisions

Governance, enablement, measurement, process maps, and designed workflows are all in place. Now decide what happens with each workflow. Automate, augment, or agentic.

See the AAA Framework below ↓

// Section 04 — The AAA Framework

Automate. Augment. Agentic.

Not everything should be automated. Not everything needs an agent. AAA is the decision tool for each workflow designed in Phase 05.

01 / Automate
Automate
Try this first
Can this workflow run end to end without a human? If the inputs are predictable, the rules are clear, and the output does not need judgment, automate it. Eliminate the thin work.
02 / Augment
Augment
When full automation is not possible
The workflow needs a human at a decision point. AI does the heavy lifting. A person makes the call. They move from doing the work to overseeing the work.
03 / Agentic
Agentic
When automation is proven and repeatable
The automation is reliable. The human decision is no longer needed. The system operates independently within the governance boundaries set in Phase 01.

The sequence matters. Start with automate. If you cannot automate the whole workflow, augment it. When automation proves itself and the human is no longer needed at the decision point, it graduates to agentic. Do not start with agentic.

As workflows move from augmented to agentic, your people move from doing the work to directing the work. That is a leadership transition. If your leaders have not been developed for it, that transition fails.

The GEM Blueprint sits alongside the LIT Framework, Become CTO, and Binary Pathway as part of a broader body of work on leadership, operating models, and modern technology adoption.

// Section 05 — Who is this for

If you are trying to lead AI well, start here.

Your board is asking about AI and you do not have a clear answer.

You need a framework that gives you something real to say. Not a slide deck. A direction.

Your team is already using AI tools but nobody is governing it.

People are experimenting. Nobody knows the cost. Nobody has set the boundaries. Nobody has defined success.

You have been told to sort out AI and you do not know where to start.

You may not be a technologist. But you still need a way to lead the work with confidence and structure.

You want to move faster without creating a mess you will later have to unwind.

That starts with governance, enablement, and measurement, not with chasing the latest tool.

// Section 06 — About

Built by a practitioner. Shared freely.

Timothy J. Hitchens
Creator of the GEM Blueprint
6x CTO ~7 yrs AWS APJ MAICD 40+ yrs 80+ orgs

Six CTO appointments managing teams across more than 10 countries. Nearly seven years at AWS across Asia Pacific and Japan. More than 80 technology organisations studied over 15 years.

The GEM Blueprint exists because the same pattern is repeating with AI: excitement first, structure later, and cost discovered too late. This framework was created to make the sequence clearer.

It is free, public, and intended to be useful on its own. No gate. No opt-in required. Just a solid starting point.

The GEM Blueprint is the public framework. If you want help applying it inside your organisation, Techshin Partners provides the implementation, leadership, and operating model work behind it. Also created Become CTO and the LIT Framework.

// Section 07 — FAQ

The questions leaders ask once the real conversation starts.

These are the questions that usually appear once people move past the hype and start thinking seriously about governance, operating risk, experimentation, and adoption.

Does governance mean people cannot experiment?
No. Governance should not shut down experimentation. It should define the boundaries for safe experimentation. In most organisations, people can explore new AI tools as long as those tools are not connected to company data, not invited into company meetings, and not authenticated into company systems.
Why allow experimentation with tools the organisation does not own?
Because experimentation is often how organisations learn what matters. Most tools do not become genuinely dangerous until they are connected to sensitive data, internal systems, or live decision making. Governance exists to manage that boundary, not to stop curiosity.
How do I know if my organisation is ready for AI?
Start with a simple question: show me your standard operating procedures. If your processes are not documented, you are not ready to automate them. And if you are not ready to automate, you are certainly not ready to make them agentic.
Why do SOPs matter so much?
Because AI works best when the process is visible. If nobody can describe the steps, the exceptions, the handoffs, and the decision points, then nobody can safely automate or redesign the workflow. What many organisations call a process is often just tribal knowledge.
Where should an organisation start: automate, augment, or agentic?
Start with automate. If a workflow can run cleanly end to end without judgment, automate it first. If human judgment is still required, move to augment. Only once the automation is proven, reliable, and properly governed should a workflow be considered for agentic operation.
Why not jump straight to agentic?
Because agentic sounds more advanced than it usually is. In practice, most organisations still need a human to own the decision, the risk, and the consequence. Courts, regulators, insurers, and boards will still look for accountable people, not autonomous software.
Who is responsible when AI makes or supports a decision?
A person is. The technology team may define guardrails, but the business must own the decision and its outcome. AI does not remove accountability. It changes where it needs to be made explicit.
Is an existing department structure enough for AI accountability?
Usually not. Saying a department is responsible is often too vague once AI is involved. Organisations need named ownership, clear escalation paths, and a record of where decision authority actually sits.
What needs to exist before something can be made agentic?
At minimum, the workflow should already be well understood, the automation should already be reliable, the relevant legislation and insurance position should be understood, and the organisation should maintain both a risk register and an AI decision register that clearly assign responsibility.
What is an AI decision register?
It is a record of where AI is making, shaping, or influencing decisions inside the organisation, who owns those decisions, what guardrails apply, and how exceptions are handled. If you cannot point to that record, accountability is still too vague.
What is the difference between a risk register and an AI decision register?
A risk register captures the exposure, likelihood, consequence, and mitigation. An AI decision register captures where decisions are being influenced or made, who owns them, and under what authority. You need both, because risk and responsibility are related but not identical.
Why do startups seem comfortable with agents while established organisations hesitate?
Startups often have fewer legacy constraints, fewer regulatory obligations, and fewer layers of accountability. Larger organisations have to think about law, insurance, governance, auditability, and consequence. That hesitation is not always fear. Sometimes it is responsibility.
Does measurement only mean ROI?
No. Measurement includes cost, time, quality, risk reduction, throughput, and whether thin work has been removed so people can spend more time on valued work. If you only measure activity or vanity output, you will miss what matters.
Who is the GEM Blueprint for?
It is for leaders, boards, executives, transformation teams, and operators who need a structured way to think about AI adoption before implementation becomes expensive, fragmented, or risky.
Is this meant to replace implementation work?
No. The GEM Blueprint is the public framework. Its job is to give people a clear and practical starting point. Implementation still requires detailed work inside the organisation, which is where Techshin Partners can help when needed.

Start with the blueprint.

The GEM Blueprint is free. Read it, share it, use it. If it opens up questions about governance, enablement, measurement, or implementation, that conversation can continue with Timothy or Techshin Partners.

Read the blueprint ↑