AI for Radical Efficiency

Most other companies are not deep into AI, despite what you see in the media hype cycle. If you feel like you're behind because of what you see happening in the news, remember that.

"You know my favorite quote, often attributed to Mark Twain, is 'the secret of getting ahead is to get started.' I think that that applies here. None of you should feel bad that you didn’t do this five years ago, because you couldn’t have done this five years ago." - Stephen Pratt, CEO

Blueprint: Get Started with AI

If you're stuck and unable to decide where AI best fits in your supply chain operation, we can break it down into four simple tasks based on ease of implementation:

  • Easy: Make current analyses better. It’s highly likely that you already have some kind of current analysis that you’re doing. You can increase your success by applying machine learning algorithms to this data.
  • Medium: Approach old problems in new ways. When you have rigid planning systems that hamper your progress because rules-based systems can’t differentiate between types of risk, it's time to incorporate new solutions.
  • Difficult (but not impossible): Tackle previously intractable problems with conflicting incentives or ideas. The idea that AI can take conflicting KPIs from different intra-organizational groups and create interconnected systems for reaching resolutions to complex and conflicting goals is appealing to anyone running a global supply chain.
  • Nuanced: Remove unconscious bias in your data. It means being vigilant about bias in your data and bias in machine learning. In order to truly eliminate bias from your data it’s important to create an ongoing process to support that goal.

You have a number that you get from whatever your process. That number is often incomplete and not as accurate as you need it to be. By moving these processes to more advanced machine learning algorithms, you’ll find that your results improve. We’ve had clients who’ve seen 5 – 10% gain over traditional methods simply by making the switch to better machine learning models.

In prediction subgroups with more intermittent demand patterns, we've seen early improvements of 10 - 20% by using improved machine learning methods. All of this improvement comes from taking the analysis you are already doing and applying more advanced machine learning algorithms and advanced data planning to your problem to create your first success.

Stop Obsessing Over Algorithms

Even the most advanced algorithm is useless without human domain expertise informing how each AI implementation advises and predicts. A good algorithm will help you do more. 

The secret's not in the algorithms, the secret's in how you solve a business problem using what the algorithms can offer you. That's the magic. 

The Best Success Comes from the Middle

Your supply chain is vast and the places where AI could benefit your business seem endless. It's so tempting to want to solve your biggest, thorniest problem first. We get it. We don't recommend it.

Potential "Sweet Spot" areas to start:

  • Predictive asset or fleet maintenance
  • Predictive quality control
  • Supply network risk mitigation
  • Demand intelligence + downstream applications
  • On-time order prediction
  • Energy prediction & shaping

Your Data, Your Foundation

You're already tracking data. It's time you put it to use. We can help.

What does "good data" look like? For starters it is in some kind of structured format - a database, for example. If it's data you are using to run your business right now, then it's useful. Depending on how advanced (or not) your data systems are already, there may be different levels of lift to get going with AI.

The Secrets to a Successful AI Project

Simple things to do to make your AI projects successful:

  • Identify a few potential use cases that touch your job, then
    get started!
  • Engage the managers who will execute and get their support
  • Define what success looks like at the outset

What can you do to set things in motion?

  • Define your keys to success - value, people, process, technology

What does "good data" look like? For starters it is in some kind of structured format - a database, for example. If it's data you are using to run your business right now, then it's useful. Depending on how advanced (or not) your data systems are already, there may be different levels of lift to get going with AI.