- September 18, 2022
- Posted by: Bernard Mallia
Prefer to listen?
If you prefer to listen to, instead of reading the text on this page, all you need to do is to put your device sound on, hit the play button on the left, sit back, relax and leave everything else to us.
Choosing the First AI Project
The emergence of AI offers leaders in every sector a chance to stand out and protect their companies. However, putting in place a company-wide AI strategy is difficult, particularly for legacy organizations.
In our view, there are three criteria for choosing your very first AI project:
The best way to find a strategic, yet manageable, AI project is to look for opportunities to automate or augment existing manual processes while figuring out where your AI strategy should take you in the long-term and framing the pilot project as the first step in that direction. Based on our experience, we are generally against the ‘start small, grow over time’ philosophy where project planning takes place in disparate silos. This will eventually create incompatibilities, will make it harder to integrate all AI areas into a single powerful engine and will end up costing significantly more.
The best candidates for AI pilot projects are usually operational in nature and can be found in any department of any organisation that has not yet embraced AI, from marketing to customer service to finance. To identify these opportunities, ask yourself: “What tasks do our employees do on a daily basis that are time-consuming, repetitive and/or error-prone?” These are good candidates for automation with AI.
Once you’ve identified a potential first project or projects, it’s important to define clear goals and metrics for success. This will help you track progress and measure ROI. For example, if you’re looking to use AI for predictive maintenance in manufacturing, a goal could be reducing downtime by x% and saving EUR y million per year in maintenance costs. If you’re looking to use chatbots for customer support, a goal could be reducing the average handle time (AHT) by z%.
Finally, all AI projects should start with an MVP – a small, focused implementation that delivers value quickly and efficiently, but is also part of a greater strategic roadmap of where the organisation wants to go with AI and where it wants AI to take the organisation in return. This is important because it allows you to learn from early successes (or failures) and iterate quickly before making major investments of time and money without going for white elephant projects that could, even if successful, hamper the growth of other follow-up AI projects. An MVP also makes it easier to get buy-in from stakeholders who may be sceptical of AI or hesitant to commit significant resources upfront without proof of concept.
When choosing your first AI project, therefore, you should look for opportunities to automate existing manual processes, define clear goals and metrics for success, start with an MVP and make sure that the project is not a stand-alone project, but fits in with an overall organisational AI roadmap, the precondition of which is obviously to have invested in such a roadmap.
To fully benefit from AI technology, your organisation’s operating context must be taken into account. Your pilot project’s primary goal is to encourage stakeholders to invest in enhancing your company’s AI capabilities, rather than exclusively create value.
The following questions need to be asked when contemplating a pilot project in AI:
- Does the project make for a quick win?
- If it does, will that quick win be perceivable as such by the decision-makers that matter within the organisation?
- Is the project too small or too big, relative to the size of your organisation?
- Is your project specific to your industry such that the benefits will be immediately clear?
In going about this task, it is important to pick a project with a high likelihood of success that can be completed quickly, preferably in under a year. Doing two or three pilot projects instead of just one will improve the likelihood of producing at least one notable success but will potentially create disruptions, delays and the realisation of cross-risks that pilot projects might not afford to have to contend with. There is no one-size-fits-all recommendation to make here, but a very delicate balance needs to be struck between finding the right project(s) and ensuring that they are managed well and without excessive disruption.
As long as your pilot project results in a rapid victory, it need not be the most important AI application in your roadmap. However, it should be significant enough that a success convinces other business executives beyond any reasonable doubt that AI is worth funding. This might admittedly create the opposite problem – that where everyone sees the value of AI and wants his/her projects to be funded first. In that case, it is important to make sure that you, as the original AI champion, retain the authority and the rationality necessary to deem how best to proceed in accordance with the AI strategy roadmap you would have drafted before the pilot.
By choosing a company-specific project, the decision makers you are trying to win over should be able to directly comprehend the worth. By way of an example, if you run a logistics services company, building an AI project in the area of recruitment to automatically screen and filter CVs is a bad idea for two reasons, namely:
Are you accelerating your pilot project with the right partners? If your AI team is still being assembled, you need to think about collaborating with external partners to onboard the required AI experience quickly. You might eventually need to establish your own internal AI team if it pays to do so. However, waiting to act until you have a team in place might be too slow given how quickly AI is advancing and how difficult it is to find AI talent.
Is your endeavour adding value? The majority of AI initiatives either enhance revenue (for instance, recommendation and prediction systems boost sales and efficiency) or decrease expenses (nearly every industry may benefit from cost reduction opportunities created by automation in accounting, call handling and regulatory compliance), or they open up new business prospects (AI enables new projects that were not hitherto possible).
Even without “big data”, which is frequently overhyped, you may still be able to add value. Indeed, choosing projects exclusively on the basis of the existence of sufficient data in the hope that the AI team will figure out how to valorise this data is a recipe for disaster.
It is not only important but fundamental to construct a specific roadmap upfront in relation to how AI will create value.