Hello everyone! Another week, more progress. I’m hoping you had a great weekend and start to the new week. This update has been uploaded after this Monday, so I want to thank you for your patience. I wanted to showcase some of the decision making of the AI, so I had to build a makeshift UI to do just that. I’m still working on the AI and trying to fine tune specific parts piece by piece. So this week I want to share some ways the AI will function in the game.
So to begin, I will present to you some images in chronological order of gameplay. The images you see have the debug UI I created to showcase this to you. The box on the left is the company’s very basic budget details that I will talk about soon. The right box is the “customer service” details for one of their Call Centers that will service this products customers. So all customer service calls and tech support calls come into this office and will be handled by their employees in that location. Just to add, you the player will be able to manage many details of your company’s satellite offices. I will talk more about that very soon in an upcoming update, but for now let’s deal with the AI’s management of their teams.
So in this first image, you can see that the company AI that I am watching is Conduit Circuitry International. They have created their newest CPU from their flagship brand the H.E.A.T Series Flame. The Flame is the name of this specific product in the H.E.A.T Series brand. Currently they just released the product and it has yet to receive a sale. Starting with the total product customers, as you can see they haven’t sold any just yet. However, the product is released and sales will be coming in soon. The total product customers simply tell you how many customers have purchased the product; not how many calls the company will get. That is determined on a few things, such as how easy the product is to use and how durable it is. These two things are the primary drivers of customer service and tech support calls. Call Productivity is how many calls the call centers employees as a whole can handle per week. So how this is calculated is based on each individual employee. They have a lot of attributes that determine how many calls they can “effectively” handle before their quality of service starts to suffer as well as their workload related happiness, which we’ll discuss in a minute.
Before continuing, lets discuss the entire customer service process briefly. Whenever you release a new product, your customer service teams need to be “oriented” or they need to learn about your product so that they can effectively service the customers. So each employee has to be taken through this orientation process which should be done while creating your product. This way, once the product is released, the team will be ready to field calls. Without orientation, they will most likely be nearly completely ineffective at servicing your customers. So after going through this orientation, they will have a score of how well they “know the product” which will govern how successful their calls will be, which governs how happy the customer will be first with your company’s customer service as well as the product. So you can see how important it is to prepare your customer service teams. Now for those of you who are concerned about micromanagement, you need not worry, because the managers you assign throughout your company can govern this. The only thing you need to really concern yourself with is how they govern which can be dictated by you… the Boss. I’ll dig more into this in another update before release.
Getting back to the AI and UI in the screens, you can also see the total number of customer service agents at the Call Center location, which is currently 58. From that number we derived the total amount of call productivity above it. So between all 58 agents, they can collectively and effectively handle 32,913 calls per week without suffering burnout or loss of effectiveness. Next is the calls processed. This is how many calls they have collectively taken this past week. At the moment, no calls have come in. Next is the workload happiness. I am still working on this one, so it is inaccurate at the moment, so for this example and update, please ignore this field.
Now you can see the percent excess employees. This will show the percent of needed or excess employees. This represents the AI’s evaluation of how many employees are needed or if there is a surplus of employees. Below that is the actual number of needed or excess employees. Finally, is the percentage of satisfied customers. Now there is still some tweaking to do, but what you’ll notice is that this number is 0% to start, it will gradually go up. Now although it is not accurate, take note that the customer service team is also “training” or orienting for the product on the fly, which means this company didn’t prepare them before releasing the product. So although this number will gradually go up, it starts low, because the team will be initially ineffective.
So now going to the second screen, a month has passed, and the team has taken 2,784 calls. The product has sold a little over 100,000 units. Now some of the numbers are from earlier evaluations in January which were conducted before the new sales came in. So at the moment the AI noticed that there were no calls coming in. As a result, the AI evaluated that there are 58 employees doing absolutely nothing! So the AI understands that to save costs, they might want to alleviate the excess salaries expense. However, the AI will evaluate again before making such a bad decision. This UI simply shows you their constant ongoing evaluations, not what they will always necessarily carry out. In this example, just think of the AI taking notes.
In the budget UI, you’ll notice, they have made some profit from the new sales. What you’ll also notice comparing this to the last image, is that the Overhead expense increased as well as the marketing expense. First, the overhead expense takes into account any added expenses for each office the company has. There are a few things that happen. First, all employees consume resources at each location, so that is water, gas, electric and so on. Each employee adds to these expensed by a small fraction. Think of it as usage cost. The more employees you have the more drain on all of your normal building resources. Secondly, the marketing expense increased, because at the end of January, the AI deemed it necessary to start more marketing campaigns for the product throughout the world. This is a 4-billion-dollar company, so it is really a small amount to spend. However, you will notice this expense continue to increase over time.
Now we’re in March and a few things have happened. Now on the right UI we see that we now are servicing a total customer base of 719,739. Out of that we’re averaging 16,537 calls a week, which might be higher as the number is literally the calls taken in last week, the next week can be significantly higher. The AI makes their decisions monthly and quarterly so that they do not often overreact to changes in the company. What has also happened is the Call Center has been downsized. I purposely programmed the AI to hire 58 agents before releasing the products to test their evaluation skills. The AI has decided that they need to cut the customer service teams down by a total of 27 employees. So when the AI made final decisions, it concluded that it could save costs and stay effective with 31 employees at this location. Now it will be some time and/or some major changes necessary before they make any new final decisions, but as you can see the AI is already taking notice that it might need 3 of those employees back if the trends in call volume continue. Lastly, you will notice that the current team finally has a good grasp on the product as they have an 88% call success rate. This rate will matter for customers deciding to purchase any of the company’s products. It will matter to some customers more than others as we spoke about in another update.
As far as the budget goes, the AI have added shipping expense as they now need to move product to stores as a result of the new demand created from their marketing campaigns. Marketing expense is down again, but will surely rise. You’ll also notice overhead is higher.
Now for the last image, there are some interesting things to take notice of. First, on the budget side, you will notice that we can now see the impact of laying off those 27 agents. It has decreased the salaries expense as a result. So of course we still needed to pay those initial 58 employees before taking them off the books. Also, marketing is back up and shipping is even higher. This is good news for the AI as it is a clear indicator of a healthy product that they have on the market. The profit margin is also nothing to shake a stick at. As for the right UI, we can see the product has reached well past 1 million customers in 3 months’ time. Lastly, the AI is now seeing that the increase in business is definitely increasing customer service calls. The AI is currently thinking they will need to add 25 new employees.
So there is a lot to take in here that is happening behind the scenes that you cannot see. First, the AI never assumes, but another company might have done things very differently. For example, the hiring manager for this company’s AI might want to keep a certain percentage of employees above that which is needed in the office where another might want to maximize costs savings like this AI for Conduit Circuitry International. You can assume that, because they laid off 27 employees within the span of a month to a month and a half period. The way it is set up, another company might have just kept the initial 58 employees in anticipation for more business. However, the difference between the two decisions is primarily cost related.
I know this update is a little technical, but I was excited to share some of the inner workings of the AI system I have created. I hope to show more soon. Thanks for reading!
