ACOS — the AI Consultant Operating System
A practical seven-phase method for taking a real business problem from first conversation to a tested, governed AI capability.
More AI tools are not automatically the answer
They need AI capability built into how they operate.
That distinction is the whole reason ACOS exists. A tool arrives, gets configured, and sits alongside the organisation. Capability changes how the organisation works — which means it touches knowledge, process, accountability and people, not just software.
For example, an organisation may need a controlled way to acknowledge common enquiries after hours. ACOS starts with that service problem, identifies which approved questions may be answered, sets the point where a person must take over, and tests the result before any wider use. It does not begin by assuming that an always-on chatbot is the answer.
Seven phases, each with a reason
ACOS is the sequence we use to understand a problem, check the information, design boundaries, build and test carefully, establish accountability, and review what changes. At every phase, you can see what happens, what we need from you, what you receive and why the phase matters.
The method is written down and used consistently. You can see the shape of the work before committing to it, understand the decision at the end of each phase and hold us to the agreed process afterwards.
What ACOS is not
| Not this | Why the difference matters |
|---|---|
| Not one-size-fits-all | The depth of work depends on the problem, risk, information and people involved. |
| Not a platform or a piece of software | There is nothing to license and nothing to log into. ACOS is how the work is done, and it is deliberately independent of any particular AI vendor or model. |
| Not a prompt library | Prompts are a detail inside one phase. A prompt library does not tell you whether the problem is worth solving, what knowledge the system should be built on, who is accountable, or how you will know it still works in six months. |
Seven phases
Each phase produces a decision, not just a document.
Discover
Understand the business, the customers, the objectives and the constraints.
Assess
Establish what is verified, what is assumed, and what has to be confirmed before anything is built on it.
Architect
Design the knowledge foundation and the boundaries the system will work within.
Build
Construct and configure the solution.
Deploy
Put it in front of the right people, in a controlled way.
Govern
Set the rules, the accountability and the testing that keep it trustworthy.
Evolve
Improve it as the organisation changes, because capability that is not maintained decays.
What ACOS considers
The phases show the order of the work. The layers are a checklist for the business, information, system, day-to-day operation, governance and ongoing review that each solution has to address.
- The business
The business itself — objectives, constraints, customers, and where value actually sits.
- The information
What the organisation knows, where it lives, how reliable it is, and what has never been written down.
- The AI system
The system itself — what it does, what it refuses to do, and what it is built on.
- The workflow
How the organisation works with it day to day, and who does what differently as a result.
- Governance
Accountability, boundaries, testing, and the rules that make it safe to rely on.
- Improvement
How it is reviewed, corrected and kept current as things change.
Governance is built in, not added later
In most AI work, governance is something written up once the system is built, usually because someone asked for it. By then the decisions that determine whether the system is safe have already been made — what it draws on, what it will answer, what it refuses, who is accountable when it is wrong.
In ACOS those decisions belong to a phase of their own, with its own outputs, and they happen before anything is put in front of people. Evolve follows it for the same reason: an AI capability that is not reviewed and maintained does not stay still, it degrades.
For retail, customer operations, aged care, healthcare and other service organisations, this is often the part of the method that matters most.
ACOS was developed by AI Architecture Lab
ACOS was developed and written by AI Architecture Lab. It is our own intellectual property rather than a licensed third-party framework, which means it is not something you can verify by looking up a standards body — and we would rather say that plainly than imply otherwise.
What we can do is show you the method in full, apply it in front of you on a small piece of work, and let you judge it on whether it holds up. That is what the Free 30-Day AI Assistant Experience is for.
See the method instead of taking our word for it
The clearest way to assess a methodology is to watch it run on something small. The Free 30-Day AI Assistant Experience is a structured evaluation built from appropriate public or supplied business information, then verified and refined with you — a controlled demonstration of the first phases of ACOS, at no cost.
Start your free 30-day experience