Inside Supply Management Magazine

Readying Your Supply Chain for Artificial Intelligence

February 24, 2020

(Editor’s note: This is the second of a two-part series on readying your supply chain for artificial intelligence (AI). This week: Learn the best practices for implementing AI. Last week’s article can be read here.)

By Ali Hasan R.

While nearly every manufacturer has at least taken steps to implement digital transformation in some form, not every company is ready for artificial intelligence (AI)-driven supply management. Organizations should certainly seek to understand the benefits that AI and machine learning can deliver, but before investing heavily in new technologies, they must first assess their state of digital readiness. This assessment involves three steps:

1) Set realistic expectations. Every organization must conduct a self-awareness test before committing to AI implementation. Gather key internal stakeholders and ask thoughtful questions that scrutinize the targets and goals of a proposed implementation.

If you haven’t yet had formal discussions about new technology integrations, decide what these integrations might help you achieve. Quantify your broad expectations for the short and long term. Weigh those against the hypothetical costs of implementation — including technology-acquisition expenses; the effects of temporary productivity disruption; and the labor costs of installation, setup and training.

At this stage, it can be useful to establish new KPIs to measure the impact of integrating AI into supply management. These should be related to the company’s traditional high-level goals. At a more granular level, professionals should understand what AI and automation would contribute to specific company operations. Digital transformation doesn’t occur in a vacuum —existing personnel and processes across the organization will be impacted, even if implementation is on a relatively small scale.

Once you have (1) an idea of the expected ROI of AI, (2) the potential impacts of digital transformation and (3) an estimate of costs, start thinking about your project timeline. Here, your focus should be on long-term efficiency gains, rather than immediate fixes. Your investment is not going to pay off right away. The benefits of AI-driven supply management are cumulative in nature, and you’ll likely have to make near-term sacrifices to achieve significant future advantages.

2) Know how the company currently uses technology. After understanding what you hope to gain from AI from a broader operational standpoint, assess your organization’s technology readiness. That assessment should be focused on three components: people, skills and tools.

Start by consulting with human resources staff to gain an understanding of the potential personnel impacts of technological transformation. Chances are good that you’ll need to bring in personnel to fill new roles in your organization, so you’ll need a plan for identifying and recruiting those people. You may also need to train existing employees and ensure they understand how their responsibilities and workflows will change during and after implementation.

Examine your existing technology stack and discuss its advantages and limitations with relevant stakeholders. Interoperability is a critical measure of tech readiness, so try to get a sense of how well your various technologies are working together now.

Do this by asking questions: Why is a language used for this application, and is it used for any others? How efficient are the data collection and storage tools, and how easy is it to retrieve data on demand? To what extent are we leveraging open-source technologies? Are our critical applications closed and dependent on vendor services and customization, or are they interoperable and application programming interface (API)-ready?

Looking ahead, you’ll also want to think about where your new tech stack will be located —on-site; in a data warehouse; in a private, hybrid or public cloud; or some combination of those. Who will need access to it (and from where) to keep operations running smoothly and KPI benchmarks met? In sum, this assessment requires a combination of meticulous planning at the personnel and application levels, and big-picture thinking about the state of the entire enterprise.

3) Dive into your data. Data is the fuel that feeds AI, and you’ll need a lot of it to maximize your returns. Most business leaders know this, and they assume that they don’t have enough data to make an AI investment worthwhile. This is a common misconception.

Within most organizations, there is usually an abundance of data being generated, stored and forgotten. For these companies, the challenge isn’t collecting new data — it’s locating, consolidating and analyzing existing data. Often, most of a company’s data is collected for compliance purposes or use during audits.

Companies will want to consolidate their business and operations data — regardless of the amount — to assess overall data readiness. And your organization probably has more data than you think. When stakeholders claim there isn’t enough data, that it isn’t clean, or that they’re unsure which data is relevant, they are succumbing to a common fallacy. They assume scarcity when availability is the real issue. Siloed data isn’t helpful to most operations, so it might as well not even exist.

Before implementing AI-driven supply management, organizations might have to spend considerable time and effort breaking down silos, which often are intertwined with company culture and deeply embedded business processes.

A lack of commonality between different personnel types, such as information technology, operations technology, and operations and business, is also a culprit. Each of these teams has a different core objective and looks at data differently. What may be immensely valuable to one department is often just noise to another, and in many organizations, a lack of regular interaction among teams leads to a lack of communication about important things like data.

Digital transformations can force internal teams to overcome silos and even restructure to facilitate increased collaboration. Ideally, however, a company should remove silos before beginning a digital transformation. Doing so will not only make the transition process easier and more effective, but provide insight on if the business is ready for such a transformation. If you can’t compel teams to work together and share important business information as a matter of course, you might not be ready.

AI is already beginning to transform the manufacturing landscape and revolutionize supply management. If your organization hasn’t yet begun a digital transformation, and you’re feeling left behind, don’t worry: Machine learning is still in its nascent stage, and AI won’t disappear. That’s not to say you should wait for AI technologies to fully mature before exploring their usefulness to your organization. Instead, follow the steps above to determine your company’s digital readiness. That exercise should inform your next steps.

If you’re not ready for transformation, prepare a plan to get ready. If you are, start creating and executing your implementation plan. The manufacturing sector is changing fast, and you can’t afford to sit still.

Ali Hasan R. is the co-founder and CEO of ThroughPut Inc., the Palo Alto, California-based artificial intelligence supply chain software provider that enables companies to detect, prioritize and alleviate dynamic operational bottlenecks.