Understanding AI Chatbots – What is an AI Copilot?

Generative Artificial Intelligence (Generative AI) is opening up opportunities to develop a new breed of apps: smart, intelligent workhorses that can do the work of hundreds of individual apps – all from a simple natural language prompt.

When you think of a copilot, the first thing that comes to mind is someone assisting a captain fly an airplane. But by the end of 2023, the word “copilot” was trending in a big way in the AI world. Take generative AI technology that we’ve come to know of recently via apps like ChatGPT and Bard and put that power right into your workflow, that is what an AI copilot is.

At a fundamental level, an AI copilot is an AI-powered assistant that can help you execute simple tasks faster than ever.

Imagine you’re about to book a business dinner with a customer in another city. Before AI copilots came along, you’d first go through the customer’s customer relationship management (CRM) data to check for any food preferences. Next, you’d open one of the table booking apps to look for a suitable restaurant to check for availability. Then, you’ll open one of the travel apps to book your travel itinerary, and, finally, you’ll open your email app to send a personalized confirmation to your customer with all the details. You’re looking at a minimum of four separate apps and at least a half hour of toil.

Now imagine this. You open one app, your AI copilot app. Instead of navigating through 4 different apps which might take several minutes or even hours, you simply type in your AI copilot app, “Book dinner with Jonathan next Monday.” Your AI copilot will work in the background and execute all of the above steps. Once done, it will send you confirmations by email and/or text, all of this with minimal intervention from you.

Beyond the evident savings in time and the obvious novelty of cutting-edge technology, it’s hard to fully convey in words the true value of this digital transformation using conventional methods. These AI copilots can do the work of dozens of apps concurrently – generate draft reports, author relevant and accurate customer service responses, compose sales emails, renew product subscriptions, pay our bills, and more. But first things first, how exactly do they get the job done?

How does an AI copilot work?

At the heart of AI copilots are building blocks referred to as copilot actions. A copilot action can refer to a single task or can include a collection of tasks required for a specific job. These may include:

• Updating a CRM record.
• Generating product descriptions from CRM data.
• Composing customer email replies.
• Handling a range of customer service use cases.
• Summarizing transcripts from chat sessions.
• Highlighting action items from meeting notes.

These tasks can be triggered via automation or on-demand in any pre-defined sequence or can be autonomously executed by the AI assistant. A copilot’s ability to understand natural language requests, work out a logical plan of action, and execute the tasks is what makes it unique. An AI assistant can handle multiple instructions (we literally mean thousands) and learn from those actions. So, the more they act, the better they get.

When multiple tasks are required to be accomplished, actions allow your AI assistant to perform a wide range of business tasks. For example, an AI copilot can help a service rep quickly resolve a case in which a customer was overbilled for a service. Or it can help a sales rep close a deal by recommending the next best actions. Want to understand in depth? Let’s get our AI copilot into action.

Take the earlier example of setting up dinner with your customer, Jonathan. If you use Einstein Copilot in Salesforce, it would know Jonathan’s initial context, like his name and CRM interaction history, but it would need a little more information from you, like date, time, and location. It could then execute actions based on your earlier one-liner instruction and respond with any other questions relevant to the associated actions: It might ask you which Jonathan you want to set up the dinner meeting with (in case of multiple contacts with the name Jonathan) and what type of cuisine Jonathan prefers if those preferences are not already there in the CRM.

What’s interesting about Einstein and other AI copilots is that they make you feel you are having a conversation with a fellow employee just like you would do over SMS or WhatsApp. But in reality, you’re just chatting with a highly sophisticated computer program. The native Salesforce SMS app serves as the conversational interface acting as a bridge between your CRM data and you and serves up information over a text conversation. The AI copilot determines what actions to execute and then generates dialogs in runtime, summarizes the output data, and paraphrases it in common human language. To you, it feels like you’re having a reasonably sophisticated chat conversation with your AI assistant. It lasts only a few seconds and then your travel itinerary is done, and your dinner is set up with minimal effort on your part.

You just tell an AI copilot – “Do so and so task” and it diligently works in the background choreographing a complex workflow of processes and rummaging through data to deliver a result that would otherwise have taken a human far more time and much more actions.

What are the different types of AI copilots?

Although the technology of artificial intelligence has been around for a while, the concept of AI copilots is fairly new. Ever chatted with a customer service rep on an app or website only to realize it was actually a bot? That’s a type of copilot. It helps customers with basic service questions but often fails to get to the deeper details of your issue. And when you get frustrated with a back-and-forth conversation that’s going nowhere, you turn to an actual human for assistance.

Chatbot technology got a shot in the arm with the launch of recent AI platforms such as ChatGPT, Bard, Google's Gemini, etc. These generative AI platforms can compose emails, write code, generate reports, and even analyze data.

With AI copilots, the interaction becomes even more sophisticated, with your own AI copilot working in the background to help you improve everything you do. The AI chatbot for Salesforce called Einstein bot is one of the several new copilot entrants in the market along with similar solutions from Microsoft and GitHub.

Here’s the key takeaway: When you are doing your research to identify an AI copilot for your business, establish one key decision parameter. Will it only use external sources for information like ChatGPT, or whether you will be able to securely connect it with all your organizational data – structured and unstructured?

Why you should use an AI Copilot

If you are reasonably well-read about the recent developments in the AI space, you would be familiar with popular large language models (LLMs) such as Google’s Gemini or OpenAI’s GPT-4. These LLMs power chatbots such as ChatGPT and are great for specific tasks. Their responses can be limited though since some of them have access to data only till 2022. And models like the ones used by ChatGPT only have access to public information about your business, they obviously don’t have access to your trusted CRM data.  Which means they can’t help you create relevant and accurate customer service replies or tell you about promising sales opportunities, nor can they act on your behalf to reply to an email or make a dinner reservation. But an AI copilot changes everything.

Let’s go back to dinner with Jonathan. Your trip was successful. Now, you may wish to thank him with a bottle of his favorite wine. Because your AI assistant already has the necessary actions to look up Jonathan’s CRM record to find his favorite brand and to charge your card on record, all you need to do is type, “Send Jonathan a bottle of his favorite wine.”

And this example is akin to the first chapter in a beginner's course on AI copilots. Imagine executing thousands of actions in virtually limitless combinations.

With an AI copilot, retail marketers can create product descriptions in multiple languages in minutes, path lab clinicians can review lab results and help doctors make diagnoses, and finance professionals can analyze mountains of data in no time to propose multiple investment opportunities. The use cases are virtually endless.

With an AI copilot, you can quickly transform your business to be more efficient and productive, regardless of the industry you work in. A conversational, generative AI-based digital assistant will do all those routine tasks that are limiting your bandwidth to scale by helping you to engage with your data like never before.

Does it seem that development around AI is happening at a breakneck pace and the very idea of wanting to figure out what you should do around AI to help your business is giving you a headache? Well, you’re not alone. As a trusted Salesforce Implementation partner for over a decade, our experts can guide you on how to combine the power of CRM, Data, and AI to propel your business into the next phase of growth.

About Author

Indranil Chakraborty

Indranil is a technology enthusiast with over 25 years of experience in project management, operations, technology and business development. Indranil has led project teams in egovernance, business process re-engineering, product development and worked with Government and Corporate customers. Indranil truly believes in the power of technology to drive productivity and growth for teams and businesses.