3. Think big but iteratively
A lot of companies are focused on running small AI projects in individual departments rather than looking at the big picture or connecting them across process areas. Ask yourself: Is focusing on how to boost productivity by a few hours per week in a single team or process as worthwhile as figuring out how to automate 80% of, say, your organization’s procurement function? If you start by thinking big and then break down the journey into incremental parts, you’re more likely to go further faster and you’ll ultimately deliver greater value.
4. Fail fast but usefully
Expect to fail sometimes—but try to do it quickly and usefully instead of spooling on it and missing out on the value you could have derived elsewhere. Likewise, accept that 90% success in a use case experiment is sometimes good enough. Given the rate of technological advance, there’s a real chance you’ll have the tools to get to 100% within the next year.
5. Take an enterprise-level approach
Rather than invest in specific AI tools for niche use cases, it’s more useful to think holistically about the technologies you should invest in. Could, for example, the AI solution being used by marketing be tweaked to perform the same task for finance or procurement? As well as reducing the cost of buying and running multiple tools, this enterprise-level approach lets you benefit from greater economies of scale and helps create better collaboration. It also reduces the risk of being locked in with certain vendors. With that said, there are cases where a very specific piece of tech may be the right answer. This is most likely to be seen with very complex industry-specific challenges.
6. Build talent as well as technology
AI engineers (a role that combines data scientist, data engineering, and software engineering into one) are and will be the most important technical talent. Yet currently the talent pool is very small. Companies can help fill the gap by investing in programs that attract new skilled employees (either straight from education or from other industries), retraining their existing technical developers, and scaling data scientists to learn engineering and engineers to learn the math. Doing so may require considerable time and resources, but it is worth it.
7. Think beyond it
Building and scaling AI-enabled systems is a job that needs multiple different skill sets beyond the IT department. This includes business subject matter experts who best understand the processes being transformed and can therefore help identify particular improvement opportunities or pain points. They are also the people most likely to use any new solution that’s put in place, so unless you involve them throughout the journey, adoption will be difficult (even resisted) and any value stymied.
Just as it has for a decade or more, data remains the backbone of modern business. Yet with the arrival of ever-more-sophisticated AI, organizations across many sectors are being challenged to evolve their approach. This means prioritizing data quality, cultivating a new generation of skilled talent and being choosy about when and where they invest.
It’s also increasingly important to look beyond small, department-led use cases and instead target enterprise-level solutions that transform data strategy across the entire company. There may be missteps and failures along the way, but by focusing on the big picture, you will be well placed to leverage its data effectively no matter where the AI era takes us. The most important takeaway, though, is to really embrace the value AI can bring and not get discouraged when value isn’t easily realized. Focus on finding which of these seven points are missing or underrepresented in your AI initiatives and address them—and the value will come.
