Why AI is indispensable to dynamically shaping company processes
Why AI is Indispensable to Dynamically Shaping Company Processes
Process descriptions are put in place to cover 80% -90% of all activities inside a company. They are intended to give guidance to new employees and to make sure that lessons learned are considered. Today, company processes are best described as fix and firm policies to be commonly applied.
All this makes sense and works up to now, but it will not be enough in the future. I am convinced that one of the biggest changes companies will have to face is speed. The speed of development, the speed of production, the speed of reacting to requests and the speed of driving information through an organization. Additionally, a crucial market differentiation will be the ability to process the non-standard requests with the same speed and due diligence as all other requests. Customers will more and more expect companies to be able to guaranty this. Therefore, it is obvious to me that the current, classical process landscape will disappear and a new approach is needed to provide a 100% coverage of all requests and, simultaneously, provide the speed and flexibility for the market.
In my opinion, this can only be accomplished by the use of Artificial Intelligence (AI) and Deep Learning. These new technologies, which are still in its infancy but coming of age quickly, will enable companies to define a few major process steps and let the AI choose the fastest and best way through the organization. Basic rules could be for example:
- check material availability before starting a production job
- products need a quality assurance protocol filled and passed before declaring the product as “ready for shipment”
- quotes can only be sent to customers if a product owner has approved them
These and many more process rules are intentionally kept general and define the principle understanding of how the company wants to make business in the market. Based on these ground rules, an Artificial Intelligence will take all the requests and lead them through the organization – individually, diligently and fast.
The next two examples describe in more detail how I think our current way of working will be affected.
Example 1
Assume a project manager receives a call from a customer. The customer wants to order a small amendment to the existing product which the project manager delivered successfully several months ago. In today’s environment, the project manager forwards the call to the responsible sales person because the project has already been closed and this is a new request/project. In an AI driven environment, the project manager opens the AI tool, selects „customer request“ and enters the information provided by the customer. While the information is being entered, the AI evaluates the inputs and responds immediately with additional questions relevant for preparing a sophisticated quote. Any question raised by the AI can instantly be passed on to the customer for clarification. At the end of the call, the AI combines all pieces to create a quote. This includes for example:
- material prices
- development effort and price
- granted discounts for this customer
- lead time calculations based on the average development time , the current load of the factory, supplier lead times and logistic processing time
In fact, the complete quote is available at the end of the call with the customer. Forget about your company target to deliver a quote within 48h. Quotes can be ready within minutes with the help of AI and Deep Learning.
Some might say that AI will never be able to estimate the development effort because every customer request is so special and so unique that a creative mind is needed. Having fifteen years of experience as a project manager, I believe that most of all customer requests can be estimated based on similar products and similar developments. These known development efforts can serve as an analogy to estimate the new request, especially, if the AI has information of all developments and all efforts spent within the whole company. In contrast, the project manager can only recall the estimates he did by himself, or he needs to contact other colleagues to gather more information. Therefore, an AI will be able to make a very good guess on the effort needed to develop the customer request – with incredible speed.
Example 2
Assume a customer order consisting of three sub-assemblies and one final assembly. Further assume the final assembly has been completed and the order is currently running through quality assurance. At this point in time the customer calls and asks for a modification. The receptionist who answers the call (the sales responsible was not available, and the system forwarded the call to him) opens the AI screen and selects the customer order. He finds the order in the system not because the customer is able to provide the 33-digit order confirmation number, the order date and the sales person’s name and date of birth. Instead he finds it by entering the customer name and a few pieces of information which popped into the mind of the customer about the order. This information is enough for the AI to select the correct order. The receptionist can instantly see that the current production is about to be completed. Knowing this, he informs the customer about the status of the order but confirms that changes are, of course, always possible.
Then, he asks the customer about details of the change request. The customer describes the change in his words and the receptionist enters this information as a „change request“ (CR). At the end of the call the receptionist confirms the CR with the customer, closes it and the production is stopped. Simultaneously, the responsible engineering department is informed about the new CR. After the CR has been evaluated by R&D, it turns out that two of the three sub-assemblies need modifications and one will remain unchanged. This information will trigger the AI to calculate a new production slot for the two sub-assemblies and a new slot for the final assembly and QA (of course it takes the estimated development time into account defined by the R&D department). A cost and effort analysis is also sent to the product manager, together with a deadline until the CR needs approval in order to meet the new production schedule. As soon as the approval to execute the CR has been given, the R&D departments modify the design. A signal from the R&D department that the design is completed will trigger the AI to order material from the supplier together with the correct construction drawings and delivery deadlines. Needless to say, the AI will check the lead times provided by the suppliers and evaluates whether these fit into the new schedule. If the R&D departments are not able to complete on time, the AI will automatically calculate a new production slot (based on the information available to the AI and further estimates based on Deep Learning algorithms).
Now, think of your own company and how a situation like this is handled today. I guess that recording the change request from the customer would have taken one week and at least required several calls with the customer. Stopping the production would, probably, have taken a couple of days – if possible at all. But most likely you would have let the production complete the final step of QA and you would have tried to convince the customer that the current product is okay for him. If you couldn’t convince the customer, you would have a big problem. You probably couldn’t salvage the one sub-assembly for the new product and your CR would be much more expensive than it should have been. Furthermore, your lead time might be screwed up because for this CR you would have to start assembly from scratch, for all modules. Most importantly, you wouldn’t be able to halt the production process for some parts of the order and would have to start the refurbishment for the other parts until all parts were on the same process step. From there, they would continue together to final assembly, QA and delivery. Without the help of AI, you would probably have to walk down personally to the production floor, discuss with the right person (of every shift) about your special customer request and, being back to your desk, you’d still have to send tons of emails to inform everybody about the situation, just to make sure nothing was scraped in the meantime.
Additionally, consider that this kind of CR is not a single event which happens once a quarter. You get CRs frequently from different customers and related to different orders. This is already real life. If you are worried, remember that the frequency will increase significantly in the future. On top of this, customers will more and more expect your company to be able to handle these last-minute changes instantly at a minimum of additional costs and time.
In my opinion, this is the challenge all companies will have to face in the next five years and it can only be dealt with if we start to use AI and Deep Learning to manage our internal processes now. We must add more flexibility and speed to our internal processes. I am aware of the fact that it is not possible to switch to these promising new technologies from one day to the other. But if our organizations do not start and make use of them now, the companies we are working in will have a hard time to survive.
I am very curious what you think about this topic. Please let me know about your opinion in the comment section and, if you find this article thrilling, feel free to like it.
Do you have any more questions?
Write a comment
Your email address will not be published. Required fields are marked with *.
