5 Facts About AI in Production Planning
Brian Hoey - October 24, 2019
Artificial Intelligence (AI) can refer to a number of things, from machine learning to computer vision, but in general the phrase is used to indicate computer programs that can reason, learn, and problem-solve from data in a way that’s reminiscent of human intelligence. This takes any number of forms, from digital personal assistants like Siri and Alexa to competitive chess playing to autonomous vehicles—each of which involves a slightly different understanding of what AI is and does. For manufacturers, the most pertinent uses of AI are likely going to be the ones that are most heavily focused on gathering insights from large quantities of data—simply because of the sheer amount of information collected and stored by most industrial and supply chain planning platforms. The question still remains, however, of what manufacturers in general and production planners in particular should expect from AI in the coming months and years.
1. It’s AI’s Logical Next Focal Point After Demand Planning
Up until this point, much of the AI implementation in industrial contexts has been in the form of improved demand planning. On some level this makes sense: companies gather demand data all the time in order to produce forecasts, and by feeding that data into a machine learning algorithm you can potentially improve the quality of those forecasts over time. Crucially, because the relevant AI is able to check its forecasts against actual demand outcomes, the software is able to self-correct more easily. When it comes to production planning, there’s less of an opportunity to do so. There’s a chance at the end of each production run to examine any disruptions or consider ways that efficiency could have been improved—but there’s no “correct answer” revealed at the end. Seemingly, this is the main reason why AI was slower in coming to production planning than demand planning; but as the technology becomes more sophisticated, production planning and supply chain planning are the logical next steps.
2. AI Can Power Continuous Improvement—If You Let It
We spoke about this a little above: one of the most powerful things about AI is its ability to power continuous improvement, which it does by learning from its own mistakes. The hitch is that it can only do that if it’s given access to clean, comprehensive operational data. In the case of production planning in particular, a successful AI deployment requires not just that your APS software collect, store, and make available relevant information about past, ongoing, and future production flows—but also that data from other departments like inventory and logistics is integrated with that APS data. If your inventory planners have their own Shadow IT that doesn’t play nicely with your APS, you run the risk that your AI will keep making the same mistakes over and over again as a result of incomplete information. For this reason, paving the way for the adoption of this technology means doing a little bit of house-keeping on your existing technology. Do you adhere to data collection and storage best practices? Does your IT system boast integration up and downstream in the supply chain? If not, you may struggle with successful AI adoption.
3. It Will Change the Role Played by Human Planners
This one might go without saying—since an AI-powered production planning solution will be able to create better-optimized plans (based on humongous caches of data) than a human planner could create on her own—but the interesting question here is: what role will human planners play instead of directly overseeing the creation of production plans? Even with the introduction of other digital technologies in the past few years, it’s by and large been necessary for production planners to provide a lot of oversight and bring their expertise to bear on computer-generated plans. As technology becomes more sophisticated, however, this need will become less acute, freeing up time for them to work on the sorts of tasks that computers don’t do well. This will obviously vary from one operation to another, but it might include working out big-picture strategic initiatives, tending to client and supplier relationships in a tactical way, or refining the integration of production planning into other supply chain activities.
4. AI Makes Real-time Planning Possible
AI is notable in part for being able to comb through and extract insights from data much more quickly than a human planner could possibly manage. Properly integrated, AI can in fact do this so quickly as to make true real-time planning possible in a way that it has never been before. Where, previously, real-time planning was made difficult or impossible by the sheer amount of data that had to be grappled with at any given second, an AI-powered production planning solution could easily power through real-time data from IoT sensors, information from RFID chips, and logistics data from elsewhere on the value chain in order to adjust plans on the fly. In this way, manufacturers can stave off disruptions before they happen.
5. It Can Reduce Costs for Manufacturers
Okay, now we come to the part that most businesses really care about: money. All of this new and exciting technology is fine—but what can it do to bring a positive return on whatever investment you make in it? Well, taking into account everything that we’ve said above, there are a few possible ways:
- Reduced disruptions on the production line
- Fewer late orders (resulting in reduced use of premium freight)
- More efficient use of resources
- Decreased inventory needs (with the potential for an increasingly lean supply stream)
- More efficient use of time for planners (enabling them to add value in other areas)
This list contains just a few of the potential areas, and it doesn’t really account for ways that the technology could actually add value (think increased product customization and anticipatory logistics), or for ways that process improvements might cascade into other areas of your operation. Still, it should give a strong suggestion of the reasons why AI is so frequently talked about. It packs the punch—and the benefits—required to eventually gain wide adoption among manufacturers.