AI supply chain optimisation for platelets to reduce costs
54%LESS EXPIRES
100%LESS AD HOC TRANSPORT
06MONTH FULL DIGITAL TRANSFORMATION
Applying ML to reducing stock wastage and increasing delivery lead times
The API Group, a large US & UK printing business came to us with a challenge that their largest customer’s forecasts were frequently dramatically inaccurate and as result caused supply chain stresses trying to fulfil their orders on time.
This customer demand unpredictability meant that they were forced to hold high inventory levels to meet orders and still they were not able to always meet delivery on time. Using Kortical’s AI cloud platform we built a highly tuned time-series machine learning model and then overlaid optimisation techniques, to make huge savings in terms of capital held in the warehouse -8.5%, while also being able to deliver on time 11% more often.
By not having good enough visibility on the future, the demand planning team were often printing and holding too much of the wrong stock and then having to work over-time to get the requested colours and styles on time. This incurred high wastage and then at times missing the agreed delivery dates, which resulted in poor customer satisfaction. They knew that they needed to employ smarter inventory management and looked to artificial intelligence to give them the ability to forecast future demand better.
They approached Kortical to do the full data science process from raw data to deployed machine learning app and even considered a UI like the below, for an easy user experience, to interact with the AI.
The first step was to build a time-series ML model, we created the lags via the SDK and then Kortical’s AutoML functionality built 10,000’s of candidate models per week, ranked each experiment on a leader board and explained each prediction all within the AI Cloud platform, to help us find the best ML solution.
Match the current delivery times and reduce the over-stock by 35% - which would have a very positive impact on cash flow but such low inventory levels it would leave them exposed to delivery risk, if there were any big changes in anticipated customer demand.
Increase the on time deliveries by 15.6% and decrease the over-stock by only 0.4% - thus API Group would be able to optimise inventory slightly and greatly improve their delivery levels.
Decrease stock levels by 8.5% and also increase on time deliveries by 11%, thus being able to get the best of both worlds where they are managing inventory optimally and also increasing customer satisfaction by delivering on time more often.
The next step was to use optimisation techniques to see the impact of either reducing overstocking or improving the delivery timeliness and finding the balance between them both.
Using a UI which shows the forecasted future demand means that the team can overlay their human intelligence to the machine learning and use whatever inventory approach suits the business objectives at the time.
Common inventory management considerations can be encoded into machine learning solutions like storage capabilities, current inventory levels, supplier lead times and schedules, seasonal trends, and future campaigns.
At Kortical we build AI solutions from your data, fully bespoke and as we build, explain and deploy them from our (Google beating) AI cloud platform we are able to do it quicker than using open source tools.
We also provide all of the tools to make it easy for you to build solutions like this yourself.
Please reach out via the form below and we will be in touch to discuss your use case and get started on building your machine learning solution for better business outcomes.
Whether you're just starting your AI journey or looking for support in improving your existing delivery capability, please reach out.
By submitting this form, I can confirm I have read and accepted Kortical's privacy policy.