It’s A Trap! Avoiding Custom AI for Supply Chain Planning

There’s loads of interest in AI for supply chain planning — but some hesitation too.

Many companies are betting that custom AI will transform their supply chain planning. Unfortunately, many of those bets aren’t paying off as organizations had hoped.

You may have noticed that there’s been a lot of talk recently about failed AI projects. Many of these projects are custom (and risky) R&D efforts.

The truth is that developing custom AI for supply chain planning is often more difficult, expensive, slower, and riskier than companies expect.

But AI does have a long track record of success in SCP applications.

Last month, we discussed 5 Things You Need to Know About AI and Supply Chain Planning.

Today, we’ll go deeper into one of the points we made in that article: purpose-built apps often outperform custom software.

As a result, buying AI-embedded supply chain planning software is often a smarter, safer, and more effective alternative to building from scratch.

This post explores the hidden costs and risks of trying to do it yourself and shows why commercial applications with embedded AI are a safer bet.

Why Use AI in the First Place?

According to McKinsey, early adopters of AI-enabled supply chain planning software have:

• Cut their logistics costs by 15%
• Decreased inventory levels by 35%
• Increased service levels by 65%

Additionally, a 2022 survey found that supply chain has the greatest potential for the highest cost savings, particularly in production, inventory management, and product distribution. By 2024, companies were realizing revenue increases from these investments.

AI’s ability to process vast amounts of real-time data also improves demand forecast accuracy, allowing companies to optimize production and inventory plans across locations and select the most cost-effective logistics solutions.

The Temptation to Build AI Software Yourself

There are several reasons your organization may be tempted to build your own AI application.

1. Belief that your supply chain challenges are unique.
2. Thinking that custom-built software will give you a competitive edge, thanks to proprietary algorithms based on your specific business model.
3. Desire for control over your tech stack and the ability to customize the software to your exact processes, while avoiding having to adapt to commercial software.
4. Pressure from internal AI/data science teams who may overestimate their ability or underestimate project complexity — especially if they’ve had success with other custom software projects.
5. Data ownership and security concerns, especially around sharing sensitive information with third-party vendors. Some organizations prefer to keep both their data and algorithms inside their own technical infrastructure.
6. Wanting to avoid becoming locked into particular vendors, giving you the flexibility to modify or replace systems as desired
7. Assumptions that internal builds will integrate better with existing systems since internal teams control all the components.

But these concerns are often given more weight than they deserve.

Why Custom AI Supply Chain Projects Fail

Building custom software is always risky, and AI isn’t easy to deploy.

That’s particularly true in supply chain planning.

What makes SCP so tricky?

Supply chain domain expertise. AI / Data Science experts often don’t have much real-world supply chain knowledge. That means companies usually lack the needed skills in advanced machine learning, supply chain optimization algorithms, and statistical forecasting methods, to name a few.

Overall expertise. Custom development teams might have strong AI capabilities but lack the resources to do all the things needed to make the software really useful — for example, to build good user interfaces, collaboration tools, workflow capabilities, and easy integration with enterprise systems.

Investments and ongoing support. Custom software can have hidden ongoing IT support costs. Companies may underestimate how much they need to invest in infrastructure, monitoring tools, security systems, and backup recovery capabilities that commercial solutions provide out of the gate.

Obsolescence. Custom solutions might meet initial needs but struggle to adapt as requirements change over time, especially if the original development team moves on.

Talent and skills. The shortage of qualified AI talent means that companies often struggle to attract and retain talent. These hurdles could lead to less than optimal results or outright failures. In addition, if team members leave, critical knowledge about the custom systems can be lost.

Data quality. Platforms are only as good as the data you give them. Continuous data cleansing, performance issues, and ongoing maintenance can prove to be a significant added burden.

Integration. Custom solutions often face significant, unexpected integration challenges with existing enterprise systems. Siloed data can be a major roadblock to realizing ROI.

Time to market. It can take years to build and optimize a custom solution. Today’s technology is evolving quickly, so there’s a risk that internal teams may build software that’s outdated before it’s even used. Commercial vendors, on the other hand, have the scale and focus to keep up with the latest advances.

So what’s a better option?

The Case for Commercial, AI-Embedded SCP Solutions

There’s a strong case for using proven applications with built-in AI from companies that specialize in building AI-powered supply chain planning applications, instead of building custom solutions.

The benefits of using SCP software with embedded AI:

1. Expertise. Embedding AI in SCP requires special skills and experience you can only get from an engineering team with know-how in both AI technology and enterprise supply chain planning applications. The result is pre-built, proven AI technology that works seamlessly in SCP applications.

2. Ongoing investment and innovation. Vendors consistently enhance their software, learning and benefitting from the economies of scale across multiple customers. They also invest heavily in R&D and have a proven track record across use cases and industries, learning from hundreds of implementations. This specialized supply chain knowledge is rarely matched in individual companies.

3. Best practices and compliance. Commercial vendors include industry best practices and regulatory compliance requirements that would be expensive and time-consuming to develop and maintain internally.

4. Scalability and support. Established software vendors offer enterprise-grade scalability, security, and support infrastructure that saves significant internal investments.

5. Reduced risk. The vendor assumes responsibility for software performance, security, and compliance, lowering your organization’s risk.

6. Cost predictability. Subscriptions and licensing models provide predictable costs compared to the uncertain budgets and timelines for custom project development.

7. Faster time to value. Companies see returns in weeks or months as opposed to the years often required for custom projects. Organizations can begin seeing ROI almost immediately with improved decision-making and efficiency.

Custom AI for SCP: Don’t Try This at Home

Building a custom AI solution for supply chain planning can be alluring, but it’s usually a high-risk, high-cost endeavor. 

For most organizations, the smartest strategy is to leverage commercial AI-embedded supply chain solutions that allow them to focus on their core business differentiation rather than rebuilding capability that already exists in mature, proven software platforms.

In a limited number of cases, a custom solution might make sense if your organization is:

• A very large enterprise with a truly unique and highly complex supply chain
• A company with strong existing AI capabilities combined with SCP domain expertise
• A situation where no existing commercial solutions address your specific needs

But in the vast majority of cases, you will maximize your chances of success if you avoid the temptation to do it yourself.

Get better results faster and less expensively by buying the right AI-embedded supply chain planning application.

To Learn More

Interested in learning how New Horizon’s AI-powered supply chain planning software can help your organization? Talk to one of our experts.