New tech platforms promise time savings and business value. They boast improved experiences that make menial tasks a breeze and increase productivity. And while some of these products provide no-brainer solutions for organizing a sales process or facilitating team communication, it’s not always a clear-cut decision to invest in a new tech platform.

When it comes to platforms that support artificial intelligence and machine learning (AI/ML), the choice is made even more difficult by the costs and complexities these solutions present. Can we trust that pre-built solutions can be purchased and achieve meaningful results? Or are we better off building them ourselves?

As a global business and technology consulting firm, we know that different organizations require different solutions. Buying solutions that automate and improve universal business practices makes sense. Supporting a company’s core and differentiated business value likely requires a custom-built solution. Additionally, there are many cases where platforms support building, customizing, and integrating solutions into a broader AI ecosystem.

Slalom’s work with a large commercial airline demonstrates how custom-built AI/ML solutions empower organizations to meet their core and differentiated business requirements. In this case, the solution brought value by optimizing operations across gates, terminals, and airports. Off-the-shelf AI solutions fail to capture the nuances and specifications of airline logistics. The implemented custom solution caters to specific outcomes that changed the game for an airline company.

AI/ML solutions have a place across business and industries; there is likely a combination of resources that is right for you and your business. There are usually trade-offs to building or buying an AI/ML solution, but clear goals and agreed-upon timelines that support the business strategy are always imperative. Before deciding where to make commitments on how to invest in buying or building, consider and evaluate your team, timeline, and goals.

In-house talent makes built solutions a possibility but requires that your organization is already invested in data scientists, analysts, and engineers. Many of the top AI/ML platforms fall short in catering to niche business strategies that apply to specific industries with unique use cases. If applying a GenAI solution to a unique, industry-specific use case is required and the talent is in place, building in-house may be the right choice. There are also considerations about deploying, monitoring, and supporting the CI/CD process around custom-built solutions, so leaders should think through the lifecycle of custom products as they weigh their options relative to their talent and in-house capabilities.

Organizations that have not or cannot invest in highly trained technical teams still have options, specifically through out-of-the-box AI/ML solutions. AutoML and citizen data science products enable the deployment of data science solutions to common cross-industry problems such as churn modeling and forecasting. However, if your application requires a level of industry expertise that is not preprogrammed into the solution, you run the risk of deploying a poor-performing model and losing out on the promised business and efficiency gains.

Buying AI/ML platforms can bring faster results, especially for standard business processes. If speed is a priority, buying a packaged solution often makes sense, but its limitations might mean waiting for industry-specific solutions. Building an AI/ML solution yourself usually provides a more customized tool set, but generally includes defining, designing, developing, and deploying a solution. And while buying can yield a solution very quickly, any failure to solve unique business cases for a specific industry can create a longer roadmap.

Define realistic outcomes. Understand how AI/ML can reasonably contribute to your organization’s unique value proposition as well as standard business processes. If necessary, seek out industry experts to help calibrate expectations. Making appropriate investments in AI/ML involves setting realistic expectations about the impact these technologies have on the business and within what time frame.

Churn modeling is a good example. Business leaders should not expect these models to produce fully accurate outcomes all the time. Instead, business stakeholders should work with technical teams to establish precision or recall thresholds that are acceptable to productively inform the business. More specifically, the team should tackle specific questions together. Is it more harmful to incorrectly flag a customer as likely to churn who was not in danger of churning? Or is it worse to classify a churner as a non-churner? Setting expectations about these business scenarios is central to setting goals and ensuring model outputs contribute to them.

Whether buying AI/ML solutions or building them, defined goals are central to ensuring AI/ML capabilities are aligned with the organization’s strategy. Standing up AI/ML capabilities that are chasing after the latest trend and not aligned to core business goals is costly because it can lead to misaligned talent acquisition and muddles the relationships between people, processes, and technology.

As with technological advancements of the past, we can expect a growing number of options for buying AI/ML platforms. Building these solutions in-house may also become increasingly feasible as organizations mature and the broader market learns to engage with this technology. Slalom experts suggest buying AI/ML solutions that don’t address niche, industry-specific use cases. If a desired outcome is specific to differentiated business value, building becomes more compelling because it generally brings more performant solutions that are worth the investment.

When it comes to AI, it’s worth doing your homework. Keep your team in mind, understand your timeline, and set realistic goals based on research and field expertise. Intentional, ongoing study of these emerging technologies will pay dividends, because although this new tech landscape can feel daunting and complex, it’s likely here to stay.