What a Fixed-Scope AI Readiness Assessment Should Deliver Before You Spend on Implementation

Many companies have already sat through an AI workshop. There were inspiring examples, a long list of ideas on sticky notes, and an agreement to "explore further". A few months later nothing has changed, because a list of ideas is not a decision, and nobody knows which idea is worth the money.

A fixed-scope AI readiness assessment exists to close that gap. It has a defined duration, a defined set of outputs, and one purpose: to tell you where automation and AI are worth applying in your business, what it will cost, and what has to be true first. This article describes what such an assessment should deliver, so you can judge any proposal, including ours, before you sign it.

Why fixed scope matters

Open-ended consulting tends to expand. Each finding becomes a reason for another phase, and the client pays for discovery without reaching a decision. A fixed scope protects both sides:

• You know the cost and the end date before you start.
• The consultant has to prioritise, because there is no time to study everything.
• The output is a document, not a relationship that only works while the consultant is in the room.

Deliverable 1: A map of how work actually happens

The assessment should start with the work as it is really done, not as the org chart or the operating manual describes it. Expect a map of your main processes that shows, for each one:

• What triggers it and what it produces.
• Who is involved and which systems they touch.
• How often it happens and how long it takes.
• Where the exceptions are, and how they are handled today.

This map is valuable even if you never automate anything, because it usually reveals duplicated work and handoffs nobody owns.

Deliverable 2: An honest review of your data

AI and automation run on data. The assessment should tell you plainly:

• Where the data for each candidate process lives.
• Whether it is digital, consistent and accessible, or scattered across spreadsheets, chat threads and paper.
• What would need to be cleaned or connected before anything can be automated.

A good assessment is willing to say "not yet" when the data is not ready, and to explain what would change that answer.

Deliverable 3: Scored use cases, not a wish list

Each candidate should be scored on the same criteria, so they can be compared. Typical criteria:

1. Frequency and volume of the work.
2. Clarity of the rules involved, versus reliance on judgement.
3. Data readiness.
4. Cost of a mistake, and whether a mistake can be reversed.
5. Dependence on other systems.

The output should be a short ranked list, typically three priorities, with the reasoning behind each. Ideas that did not make the list should be recorded too, with the reason, so they are not rediscovered from scratch next year.

Deliverable 4: A recommended path for each priority

For every priority use case, the assessment should recommend how to deliver it, because the right answer differs:

• An existing product when the need is common, such as CRM, bookkeeping, payroll or customer messaging. For these, a business operating system such as Polynusa Autopilot is often faster than building.
• Integration or custom software when the process is specific to you, or when existing systems must stay and be connected. This is the work of Enterprise Software.
• Devices and sensors when the data has to come from the physical world, such as equipment, energy or buildings.
• Process change first when the problem is unclear rules rather than missing technology.

Deliverable 5: Cost estimates you can test

Each recommended path needs an estimate of cost and time, including the parts that are easy to forget: data preparation, integration, training, and maintenance after launch. The assumptions behind each estimate should be written down, so that you can compare them with vendor quotes and check them later.

Deliverable 6: Risks, prerequisites and how success will be measured

Finally, the assessment should list:

• Prerequisites, such as data that must be cleaned or an owner who must be appointed.
• Risks, including where human review must stay in place.
• Success measures for each priority, with the baseline recorded now, so that results after implementation can be proven rather than claimed.

How to judge an assessment proposal

Before engaging anyone, ask:

1. What exactly will I receive at the end, and in what form?
2. How long will it take, and what does it cost in total?
3. Whom in my team will you need to interview, and for how long?
4. Will the recommendations be independent of the products you sell?
5. What happens to the fee if I proceed to implementation?

Clear answers to these questions are a good sign. Vague answers usually mean the scope will grow.

It also helps to ask for an example of a previous output, with client details removed. A sample tells you more about the depth of the work than any slide about methodology, and it shows whether the final document is something your team could actually use to make a decision.

How Etzal Group runs it

AI Consulting at Etzal Group is a fixed-duration package with a written output and a cost estimate. It maps where automation and systems are worth applying in your business. If you proceed to one of our implementation lines, Polynusa Autopilot, Enterprise Software, or IoT and Control, the consulting fee is deducted. To discuss your situation first, use the consultation page.