It is not a secret that technology evolves at an accelerated rate. As an example, I have seen the progress of vinyl records to cassettes to compact discs to playable digital files and finally to streaming media. My teenage sons, on the other hand, have never owned a CD and barely recall downloading mp3s.
One of the more interesting advances in technology, particularly in a business and consumer setting is in the area of Artificial Intelligence (AI) and Machine Learning (ML). Building upon the confluence of device (sensor) proliferation, high speed data ingestion and advances in computing power, AI/ML produces use case-driven insights that we leverage in what seems like all aspects of our lives.
It is an advancing space that is both obvious – think of the personal assistant on the smart phone of your choice – and yet subtle – think of the fitness device insights and streaming service viewing suggestions – at the same time.
In the world of logistics, the contributing forces are similar. The availability of disparate data and highly scalable ingestion engines, the advent of performant and elastic public clouds, and the wider adoption of open application program interfaces (APIs), presents a ripe opportunity to apply AI/ML to provide unique, incremental value.
But like many things, how do you separate hype from reality? In our consumer world, if our streaming service suggests something that we wouldn’t watch, we just skip over it. If my fitness watch tells me to rest when my coach tells me to train, then I train. In the world of supply chain, however, poor or weak insights have real impact and real cost.
To separate the wheat from the chaff, I suggest considering three approaches:
- Avoid the general: Many solutions will boast being “AI/ML powered” as if it is just something you turn on and it magically makes things better but that is not how it works. Learning models are designed to focus on specific cases. Labor and order forecasting, and transportation risk assessments are good examples of particular use cases.
- The right data in abundance: AI/ML models require data and lots of it. The good news is that in today’s age, data is abundant. The more difficult question is understanding what data the model is trained to ingest and leverage and how reliable are those data sources. A model to evaluate execution risk on planned loads will be limited if the scope and availability of constraint and network data are overly simple or unreliable.
- Proven in reality: It is easy to come up with a premise of value, but a premise is immaterial unless it is tested with real data ingestion and measured against a control. Only by creating real world circumstances can a premise be proven and more importantly measured.
Applying this in practice, let’s take an example of a logistics solution promising AI/ML powered routing. Working through the steps:
- Avoid the general: What aspect of route planning is being improved? e.g. load fill rate, driver assignment, route sequence.
- The right data in abundance: What data is the model leveraging? What is the timeliness and reliability of that data? What is the scope of the data being considered?
- Proven in reality: Has the problem been solved in a real-world scenario? In what circumstances e.g. a grocery environment will have different data and considerations than an industrial manufacturer.
The acceleration of technology is enough to give anyone pause. The opportunity these advancements present are unprecedented but also forces us to really understand what data is there. By interrogating three simple perspectives, we can quickly separate the real from the noise and take our logistics networks from vinyl to streaming.
About the Author
Fabrizio Brasca, Group Vice President, Global Solutions, Blue Yonder
Fabrizio (Fab) Brasca is group vice president (GVP), global solutions at Blue Yonder. In this role, he leads his organization to drive thought leadership, go-to-market strategy and solution execution excellence for Blue Yonder’s 4,000+ customer base. Additionally, his organization works closely with customers to understand requirements, drive best practices and adoption of Blue Yonder solutions across the globe.
Previously, he was VP of solution strategy, supply chain execution, responsible for developing innovative strategies across all industry verticals, strengthening executive-level relationships with JDA’s key customers and prospects and advising companies on best practices to become more profitable. Fab joined Blue Yonder as part of the i2 Technologies acquisition in January 2010 after spending more than 10 years at i2 serving in transportation-focused senior management positions. In these roles, he helped lead i2’s global transportation practice including marketing, presales, roadmap development and services functions.
He holds an Honours Bachelor of Mathematics, with a specialization in business and information systems from the University of Waterloo in Canada.