How You Can Create Agent Schedules In A Few Minutes With A Custom GPT

As a contact center manager, you know that manually creating agent schedules is tedious and time-consuming. The alternative is to automate using a Workforce Management (WFM) solution, which can be cost-prohibitive if you aren't using it to its full capacity.

Having been a contact center manager myself in my previous career, I have spent countless hours pouring over Excel sheets to ensure that I not only accommodate all the scheduling restraints of the contact centers, but also give every agent fair hours across the schedule. So, I can certainly sympathize with you! 

In this article, I want to show you a simple yet effective way to use OpenAI's ChatGPT (the Custom GPT function, to be more precise) to help you create a schedule you can tweak in minutes. Don't worry; although there is a little upfront setup involved, you don't have to be technical at all to do it.  

Why Contact Center Agent Scheduling Is Hard

Creating schedules for your agents manually is challenging and tedious work that often takes forever because schedule management is highly complex. Contact centers usually deal with various scheduling constraints, such as peak hours, agent availability, and skill requirements. This means to create a schedule, you have to balance multiple variables. Finally, it is complicated to adapt the schedule to sudden changes, such as unexpected absences or shifts in call volume.

On the flip side, the reasons that manual scheduling is difficult make it an ideal use case for Artificial Intelligence (AI). AI can automate tedious, repetitive scheduling, saving time and reducing human error. In addition, AI-driven scheduling can be more responsive to these fluctuations, ensuring better coverage.

What Is A Custom GPT & How Do You Create It

A Custom GPT is a tailored version of OpenAI's GPT-4 model, customized for specific tasks, industries, or applications. It allows you to adapt the Generative AI model to better fit your needs by fine-tuning its responses, personality, and behavior according to your requirements. This might sound complex, but it is super easy.

To show you what the end result can look like, I created a GPT called Agent Scheduler. You can access it here (you will need to have a ChatGPT login) and play around with it. This will give you an idea of what you can expect from a GPT. Keep in mind that this is a sample and you will need to train yours on your requirements so that your schedules meet those criteria.

Imagine you just hired a very helpful and enthusiastic intern to help you create schedules every two weeks. First, you will need to train your intern by explaining to them the scheduling constraints of your contact center and share with them all the helpful information they need to know in order to create a great schedule. This is exactly what you need to do as you create your Custom GPT. If you have the paid version of ChatGPT, here is how to set up your own:

1. Create A GPT

Image: Screenshot of ChatGPT interface. This screenshot shows you what to expect when you create a new GPT. Click "Configure" to follow the steps below.

2. Customize Your GPT

Click on "Configure." Now you can set it up by giving it a name and brief description. This is what you will see every time you open your GPT as a user later on. If you make your GPT publicly available, everyone else will be able to see this too, so make sure it is descriptive and accurate.

Now, in the instruction section, you will tell your GPT what to do. Talk to it like you would to the intern mentioned above. Be aware of the curse of knowledge: just because you know something doesn't mean your intern/GPT knows it, so be very clear, specific, and comprehensive. Include everything you think is important to know. This might even include how it should respond to queries, its tone, or other behavioral aspects.

In my example GPT, I gave it this simple instruction:

"You are a Contact Center director in charge of making biweekly schedules for its contact center agents. Here are some things you'll need to know in order to make the schedule:

• Our agents need to work a minimum of 20 hours per week.
• Their daily shifts will alternate between four, six, and eight hours in length.
• Our agents need at least one day off per week.
• Our operating hours are between 08:00 a.m. and 08:00 p.m. Monday through Friday."

Try to get as detailed as needed while remaining concise. 

Finally, I highly recommend that you upload one or two schedules you have created in the past in the Knowledge section. This will show your GPT what kind of output you are looking for and what good looks like. 

Now your GPT is ready for testing.

3. Testing & Tweaking Your GPT

Save and publish your GPT (I recommend keeping it private) and then try it out. Remember: just like an intern, your GPT requires a bit of training to get it just right. 

Try using a simple question like, "Hi, can you please create a bi-weekly schedule for me?"

Below is the schedule I generated from my sample GPT. If you try it out, simply click on the two expanding arrows on the table to get the full view of the table. You can also download this schedule into Excel by clicking on the downward arrow with a line on the bottom.

Image: Screenshot of my sample Agent Scheduler GPT. In this screenshot, you can see the expanded schedule it created based on my input. Pretty impressive.

See what happens if you add a starting date. Try "Hi, can you create a bi-weekly schedule for me? The first day of the schedule should be August 26th, 2024."

Or you can ask the GPT to add a calculation of hours worked per day or per agent in the last row and column respectively. Try to ask one question at a time to see what happens. 

It is really important to understand that the better you train your GPT, the better the outcome will be. So keep tweaking it (just click edit GPT) and in no time, you'll get usable results you can download in an Excel Sheet. 

Image: Screenshot of my sample Agent Scheduler GPT. As you can see, I can give it follow-up instructions, like calculate the total number of hours worked per day — which it did here nicely. It did not delete row 56 but that's okay. I can fix this easily in Excel.

Prerequisites

There are a few prerequisites that you must have in place before being able to set up your GPT. 

• Historical Data: To train AI models effectively, you will need a sufficient amount of historical data, such as schedules you have created and used successfully before.
• Defined Objectives and Constraints: For the AI to create a usable schedule, you need to have a crystal clear understanding of your contact center scheduling restraints and your desired business goals, e.g., service levels, etc. 
• Agent Profiles: Detailed profiles of agents, including their skills, availability, and preferences, should be available to personalize scheduling.

Once you have these things in place, you are all set to start creating your own GPT Scheduling Manager.

Are You Ready To Get Started?

As you can see from the instructions above, the initial setup takes no more than a few minutes. You will need to test and tweak it a little to get it just right for your requirements, but within an hour you should be ready to go. 

Congratulations! You are now on your way to saving hours every week. Even if you have to sanity-check and make a few tweaks here and there, you will be able to eliminate this tedious and time-eating task from your to-do list from now on. 

Let me know what you think of the GPT and how far you get on creating your own. Was this helpful? Would you like to see more like this? Let me know in the comments.