AI for real work · 2 of 5 · Choose
Which AI project should you start with? Impact, complexity and risk
One good idea invites action. Several good ideas require a choice. As you recognise opportunities to apply AI, you may find more worthwhile initiatives than your business can develop at the same time.
Choosing commits time, attention and resources. It also determines what the team will learn from its first experience. Setting a priority therefore deserves more than enthusiasm for a tool or a quick impression of its benefits.
In the previous article, we began connecting a difficulty in the work to a possible improvement. We will now use that connection to decide where to begin.
Understand what impact means
Comparing opportunities requires a shared reference point. That reference is the purpose of the business and the importance of the change we want to make.
Impact describes how much an improvement would contribute to that purpose. It may involve quality, time, capacity, continuity or revenue. Evaluating it means explaining how a change in the work would lead to that result.
Frequency and reach matter, as do the effects on the rest of the activity. A difficulty may take little time yet influence important decisions. Another may recur often enough to accumulate considerable effort.
Understanding these relationships helps us locate the value. It also makes it possible to discuss it with people who know the work and can judge whether the improvement would matter. An estimate becomes better grounded when we can explain the observations behind it.
Ask what achieving it requires
Once we understand the possible value, we need to consider the effort required to achieve it. This is where complexity enters the discussion.
Complexity comes from several connected conditions: what we know about the problem, the information available, the variety of situations and the links to other systems or people. We must also account for verifying results and handling exceptions.
When these conditions are poorly defined, part of the initial effort is to understand them. Recognising that uncertainty helps us distinguish an idea that is easy to explain from a solution that is easy to put into practice.
Comparing impact and complexity gives us an initial map. We can identify opportunities that justify an early test and others that need preparation or a narrower scope.
Consider consequences and timing
The map is still incomplete. Two initiatives with similar value and effort can have very different consequences if something goes wrong.
We need to understand who will use the result, which decisions depend on it and how an error would be detected. These connections determine the review required and help us estimate its cost. They may also reveal a condition that must be resolved before starting.
Timing matters for a related reason. Some opportunities retain their value while we wait. Others depend on information, people or conditions whose availability may change. Urgency becomes meaningful when we can explain what would be lost by postponing action.
I want those differences to remain visible. A score helps organise the comparison; an explanation makes it possible to judge it. When assigning scores to impact, complexity or risk, we should retain the observations behind them and identify where we are still making assumptions.
Choose what you want to learn
A first project can produce another valuable result: evidence that improves future decisions.
We can ask which important uncertainty each initiative would resolve. We may need to know whether information is sufficient, whether a technical capability reaches the required quality or whether people find the new way of working useful. Understanding this can change our assessment of other opportunities.
That learning becomes more valuable when the test is bounded and its conclusions can be interpreted. We need to connect what we observe to the conditions in which it happened. The first project can then offer both a possible improvement and a firmer foundation for later decisions.
The AWS guidance on AI adoption relates the exploration of opportunities to business objectives and capabilities. Here we add a practical question: which first step lets us move forward and learn with the resources available?
Narrow the scope to make a first step possible
The most valuable opportunity may initially seem too broad. Before setting it aside, we can examine whether it contains a part we can investigate on its own.
Defining a scope means specifying the work we will cover, the conditions involved and the expected result. That boundary must retain enough importance for the test to matter and enough simplicity for it to be feasible.
A narrower scope also makes dependencies visible. We can distinguish requirements that need attention now from capabilities that can be added later. A large initiative can then become a sequence of manageable decisions.
The first choice should lead to a clear priority: what we will explore, why it deserves attention, who will participate and what we need to discover. Other ideas can be kept together with the reasons for revisiting them.
Choosing in this way turns a list of possibilities into a beginning. The next step is to check whether the conditions are in place to pursue it.