As you plan next year’s strategic objectives, you are, like your competitors, focused on optimizing operations and freeing up staff to concentrate on value-driving activities. AI seems promising, but where do you start?
Asking this question is a positive first step. Many organizations begin their AI journey with a solution and then try to fit it to their business, which can be risky.
Stakes are High
While the above scenario may sound obvious to avoid, the implementation and operation of AI solutions present some unique challenges that make project failure particularly impactful to organizations.
- Cost: Given the data prep required, compute and storage infrastructure needed, complexity of integrations, and ongoing model maintenance. Implementing and operating AI solutions can be more expensive than other IT solutions.
- Organizational Change: AI solutions transform how people perform their jobs making resistance to adoption a major factor without the right training and change management strategy in place.
- Data Privacy & Compliance: Mishandling of sensitive data has the potential to cripple an organization’s ability to operate, particularly in regulated industries.
The good news is that these factors can be mitigated; and in every case this begins with how you start your project.
The Right Business Problem
It is critical that the return on your AI investment exceeds the level of risk you will be taking on. This makes identifying the right business problem paramount to the entire process —and as you’ve likely experienced before —can be easier said than done.
With the speed of change these days, knowledge silos build up quickly in organizations. While these are often necessary to operate your business efficiently, they can be problematic when attempting to identify the highest leverage business problems in your organization, which require a more holistic understanding of how your organization operates.
Compound this with the number of steps, handoffs, dependencies, inputs, and outputs involved in any given operation and you have a full-blown project that needs to be completed before considering what AI solution is appropriate for your business.
Documenting Your Operation
It is common for companies to have difficulty maintaining their process documentation let alone creating it to begin with. However, when it comes to implementing AI, having current documentation of your operation is not optional. To identify the right business problem, you need to be able to visualize the inefficiencies in your processes.
A strong Business Analyst will be able to conduct the necessary interviews and put together a visual that will succinctly show all processes, teams/resources, systems, and inefficiencies within your operation, enabling you to target the right business problem in your organization.
Defining Your AI Use Case
Once you’ve identified your target operation(s), continue to leverage your Business Analyst to define your AI use case. Your use case is the artifact that will garnish support for your AI initiative and ensuring it is a worthy investment. At the tactical level, a strong use case will articulate the business problem, explain its impact to the organization, highlight the benefits, and define what success looks like. Make sure to define your use case before selecting an AI tool to avoid solutioning before fully understanding the problem and goals.
To make things more concrete for stakeholders, there is value in exploring how your AI initiative will come together. While these factors will be variable and not understood until later in the process, creating some initial assumptions early will help in creating a holistic approach to your initiative.
- Create a list of dependencies or precursor projects.
- Draft a support plan for maintaining the AI solution.
- Provide a high-level budget estimate.
- Identify KPIs that will be used to measure success.
- Establish a Change Management strategy.
Other Considerations & What’s Next
We have engaged with a number of clients who were interested in starting their own AI initiatives but after further analysis determined that an automation solution would be a better fit for their needs and more cost effective.
Before you start scoping your AI initiative ask yourself this series of questions:
- Is the operation rule-based and repetitive?
- Are the inputs and outputs for the operation predictable?
- Is consistency in outputs more important than adaptation?
- Can this operation be performed with minimal amount of thinking?
If your answer to all or most of the above questions is “Yes” you’re likely looking at a problem that can be solved with automation. In comparison, AI solutions are best suited for operations that require decisions to be made under uncertain or changing circumstances, outputs are more predictive than execution-based, and inputs into the operation should inform/evolve future outputs.
What’s Next
After identifying your business problem and specifying the use case, the next step in implementing an AI initiative is to address data readiness, which is a fundamental requirement for any AI solution. For further details, see Why Data Governance is the Foundation of Successful AI Initiatives.
To learn more about this subject and other considerations when preparing your organization for AI, consult AIM’s Playbook for AI Success.
Insights By
