AI came roaring into the supply chain’s mainstream consciousness in the last 2 years.

There was a clear inflection point in research achieved when large language models began quickly churning out near human-like responses to prompts. It gained more fervor when the first image-generation tools came out. 

A generational leap for AI supply chain tools

Aside from some gibberish (not considering accuracy, just readability), and the occasional sixth finger, the results were hard to argue with. We had achieved a generational leap forward in computer-generated content.

The genius of making AI tools publicly accessible went a long way. A lot of innovation goes on behind closed doors, in lab settings. This was, essentially, a public beta test. Everyone was invited to play with it, build on it, and gain emotional investment in it.

Supply chain applications quietly chug along

Beyond some personal assistants and content summation tools, there haven’t been things that addressed the Operations teams. When it comes to the true backbone of most businesses – the acquisition and distribution of physical goods, we’re waiting for the major move forward.

AI supply chain disillusionment

To borrow from Gartner, we’re heading into, or firmly in the trough of, AI disillusionment right now. We’ve been sold the promise, we’re excited for it, but we’re not feeling the benefits of AI in the supply chain. 

It is important to remember, right after the trough of disillusionment is adoption and productivity. Big breakthroughs are on the horizon for the supply chain.

Remember, AI isn’t new to the supply chain, only the way we popularly think about it. Business intelligence, advanced analytics, and recently data science have always found a ready audience with supply chain professionals.

AI is evolution, not revolution

Many of the techniques used in supply chain advanced analytics – from decision trees to optimization algorithms – are, in the simplest terms, AI. 

These techniques produce a conclusion based on inputs and can shift depending on data points, records, or sets of records that are fed to the systems such as the warehouse management system (WMS). They replicate, codify, and replace the way a human would reach a conclusion under the same conditions. Some are simple; some are incredibly complex. But, AI is Artificial Intelligence. In this context, we’ve deployed AI-based solutions for a long time, in simple ways.

It would be lax to not address that future AI supply chains are not a revolution or an instant leap forward. Rather, they are an evolution of these foundations. 

What changed?

AI in the supply chain provides enhanced availability of public data sets and improved free (or cheaply licensed) packages have provided the tools needed to drive the next wave of innovation. Coupled with economical access to cloud computing and on-demand virtual CPU and memory assets, AI suddenly had access to substantially more power, significantly cheaper, than any previous generation. In simplest terms, horsepower and fuel were no longer limiters. 

The ‘cost per unit’ is finally hitting the point where it can be used en masse.

What does the supply chain need from AI?

Some of the humorous issues with aggregation AI can make the strategy unsettling for logistics-focused businesses. 

You can’t trust an automated process that doesn’t reach the same outcome using the same variables every time. The core, boring, day-to-day should be predictable, consistent, and repeatable. You aren’t looking for an impressionist shipping label and your packing slips should be boring and dependable.

The desired outcome of AI in the supply chain is predictability. What that looks like depends on who you’re talking to – whether that’s supply chain Executives, Operations, or IT. 

The biggest pain that supply chains feel, regardless of role or industry, is the value-add use of time and the constant pressure for efficiency. Anything that can’t be billed is ‘waste’ in the classic 8 wastes sense. Thus, the AI industry has to support that core mission.

So, where will the big AI wins come from?

  • Root causes analysis: Anything that goes beyond an obvious cause-effect is tedious or defies research. Multi-variate analysis that can string together hidden causative associations can remove the need for human research, as well as take the results and prescribe corrective actions. It also doesn’t rely on a human to respond to the flashing light; the process is kicked off automatically. You just deal with the solution, not figure out the problem.
  • Active monitoring: The best fix is to prevent the problem in the first place. The root cause techniques can be used to proactively consider the business and raise notifications for things that match previously experienced patterns of failure or create suggestions for smoke detectors.
  • Configuration suggestions: So many legacy fulfillment systems were based on set-and-forget, locked into the use case during implementation. If the market changes, do your settings update? Do you consistently produce old answers for new problems? AI-based supply chain solutions can consider if trends are changing in the business and if your rules should be updated for the new normal.
  • Advanced, dynamic forecasting: Simple demand forecasting has been around for a long time, but the market is finding that the dynamic nature of global business as well as the acceleration of business cycles (or the complete disappearance of them) is breaking those rudimentary approaches. The need to consume causal data, sift for correlation, and then do it again, will see a strong adoption of AI.
  • Copilots / personal assistants: Many are leery of fully delegating control to the black box. This makes the ‘AI assistant’ approach much more palatable. The ability to deploy a bounded AI that seeks opportunity or reads tea leaves for pain points and drives the user to respond can quickly alleviate the manpower as well as experience issues in our space. Minimizing the gap between experienced and new operators will drive productivity and consistency. One of the strongest use cases has been distillation, summation, and bullet point creation.

Think about who’s impacted

Any use case for AI that minimizes manual intervention while improving intelligence quality is an immediate win for the supply chain. 

Not only does this keep your people focused on execution; but it also improves the speed and consistency of any exception-based activities. Continuous improvement focuses on the identification of non-value-added actions; AI presents our most powerful supply chain tool for minimizing and automating those events.

Not everyone is going to feel the benefits of AI supply chain tools – and risks – the same. So it’s good to think about who’s impacted.

To each their own

All businesses will, to one degree or another, consider Sales as their primary function. A supply chain is a way to fulfill that demand. This means that when considering something as disruptive as AI in fulfillment, we’ll want to consider 3 core personas that have specific needs to satisfy.

What AI means to supply chain executives

The leadership team struggles with needing answers for exceptions. But they also generally struggle with having their attention pulled between multiple functions. Supply chain leaders need to be made aware of issues and opportunities quickly, so they can make informed decisions. 

Often Operations can feel like the Executive is informed on pain points, but then isn’t engaged when things are working – or slipping. This is often when subtle smoke signals could head off problems if they were easier to sense.

The time between researching a problem and actioning on it often will eclipse the window of opportunity. If no information is frustrating, then information as a post-mortem is even more frustrating. 

AI supply chain tools provide executives with timely knowledge so that something can be done. 

  • AI analysis tools: constantly monitor the business so they can take action immediately without needing to pull a fire team together.
  • Visibility and pre-warning: these are absent from most products today. Escalation is manual and based on the awareness of the floor team. The length of wait between an incident, awareness, and escalation exacerbates the first point. 
  • Trigger- and threshold-style awareness: improve capital expenditure planning by making analysis a continuous business strength rather than responding after everything’s broken.
  • Fewer surprises: executives want AI in the supply chain because it keeps them closer to the floor and proactively informed, leading to better directional decisions.

What does AI mean to supply chain Operations?

The supply chain operations team is concerned with executing predictably. 

AI, by design, is not predictable. While it is easy to make the use case, the standard pushback will be, “We don’t have time for today’s work. A new process would be too disruptive.” This is compounded by an inherent distrust of any ‘black box’ technology whose function is not obvious.

In addition, operational concerns for labor retention and replacing people with computers will increase resistance to adoption or even bringing on technology.

AI supply chain tools provide Operations with the chance to make their workers’ jobs easier.

  • Faster, easier decision-making: There is a strong need to remove decisions that aren’t decisions. Detection of workflows that always take one route through a decision tree can be codified into recommendations and presented to the worker.
    • Simulations and predictability: Highlight configuration changes based on changes in either the customer order profiles or how orders are processed. This will be raised based on target KPIs that the system can simulate changes to achieve. These rule changes will be presented to the user so there are no sudden changes in the way the system works. Trust, but verify, and accept.
  • Time savings: Expedite root cause analysis. Spend less time hunting through data and tracing workflows and focus on remediation and serving the customer. AI is uniquely structured to identify unintuitive casuals within the facility that might be hidden in a strictly documentary search.

AI is not in the supply chain to eliminate jobs. Most environments have a lot of labor that is tasked with error identification, elimination, and rework. But those aren’t “value adds” – they aren’t things the customer wants to pay for. 

Hours freed up from fixing problems can be reallocated to serving the customer and growth projects. Few operations need less staff. They need staff that is more effectively deployed and informationally served.

What does AI mean to IT and Security?

Most IT functions focus on the adoption and deployment of tried-and-true solutions. The book is still being written on AI best practices, so a high degree of subject matter ignorance and a function steeped in risk aversion will mean many AI use cases/projects will be ignored or actively fought.

There are several benefits of AI supply chain solutions to the IT leader, but caution needs to lean toward protecting the business from bad practices and worse actors.

  • Deploy public or private AI models: Public Large Language Models (LLMs) are reliant on a combination of provided information and publicly scrapable data. Private models are NOT dependent on access to the public domain or copyrighted materials. Know what you’re deploying.
  • Protect your data: It must not be fed to the Internet and used to feed the public algorithms such as what happened with Microsoft’s AI GitHub repository. Your data has value – to you and your firm’s clients – don’t give that away. Choose an AI partner that doesn’t have clauses claiming your prompts and data.
  • Let the software do the work: You aren’t acquiring pricey data science teams and maintaining one-off projects internally to harness these advancements; you can let the software provider own that.
  • Productized AI benefits: No need to build it yourself. Choose a platform built from work with an entire customer base as opposed to some niche, bespoke product developed for a one-off. Thoughtful and proven; not reactive and risk-laden. 

Unlike past advancements, the deployment of AI in supply chain solutions and the business isn’t a one-person, one-organization decision. It has to be cautiously approached, but the benefits are enormous if the appropriate protections are in place.

Deposco is your most valuable operator

Deposco doesn’t view AI as just chatbots.

Every software company in the world in trying to create a little chatbot on their help site. That’s all fine and good, but not the big hit advancement we’ll see in supply chains.

At Deposco, our AI investments address a key problem: business units driving the supply chain are too disjointed and gut-driven in their decision-making process. IDC cited key factors contributing to decision-making challenges — including the number of variables to consider when making decisions, lack of access to the required data, difficulties integrating the necessary technology, and more:

33% of decisions are made primarily based on intuition and experience and 25% of decisions that should be made are not, IDC revealed.

It’s about not making users dig, and getting better at what they do

To enable our customers to make timely, sound decisions and keep up in dynamic environments, Deposco provides a single, end-to-end platform that delivers actionable insights and curated content for continuous improvement. 

AI-driven analysis in Deposco allows your team to get recommendations and curated content that include auto-generated explanatory analysis about why an event happened, and the causes that made a key metric trend in one direction. So users don’t have to dig.

  • Personal data concierge: Get AI-powered market insights on how your key operations metrics like pick lines per hour measure up to comparable companies in the market that are in the same or similar industries and handle similar product types. Gamify metrics improvements to create a virtuous process improvement cycle for customers and internally.
  • Efficiently & accurately examine what works or doesn’t. Use AI-based decisions set to automatically fine-tune your processes in the background. Not a set-and-forget system… Auto adaptable, automate, optimize, reevaluate, and adjust.
  • Constantly evaluate how certain system parameters are tuned. Deposco learns the performance of the output from those tuned parameters and then auto-adapts and applies changes to those parameters to improve results for future process runs, thereby continuously optimizing operations without requiring a lot of expensive manual intervention.