Intro to AI and How It Impacts Your Salesforce Strategy (Part 1)

If you are part of the Salesforce ecosystem, you’ve undoubtedly found yourself caught in the whirlwind of Artificial Intelligence (AI),

5 min. read

If you are part of the Salesforce ecosystem, you’ve undoubtedly found yourself caught in the whirlwind of Artificial Intelligence (AI), and you’ve likely been told that you need to start preparing your organization to harness the power of AI. I completely agree that you do need to consider the strategic role that AI plays in your business. However, the sheer volume of information out there on what AI is, its capabilities, and which Salesforce features are crucial can be a daunting maze to navigate. This, in turn, can present a significant challenge when it comes to crafting an AI strategy for your organization.

With that said, this is the first post in my “Introduction to AI and How It Impacts Your Salesforce Strategy” series. Over the course of this series, I plan to provide an introduction to AI, explore the different types of AI and their practical applications, and highlight the various areas of the Salesforce platform where you can leverage AI-driven features.

My hope is that this series will assist organizations that are at the onset of defining their AI strategy in navigating the wealth of information available on AI and Salesforce, helping them determine which features will offer the most value for their organization and outlining the necessary steps to position themselves effectively for feature utilization.

What is AI?

My favorite definition of AI comes from Salesforce’s “Get Started with Artificial Intelligence” trailhead module. It defines AI as “as the ability for a computer to perform skills typically associated with human intuition, inference, and reasoning”.

If we boil AI down to using a computer to perform certain skills typically associated with human intuition, there are two key components to this:

  1. Teaching (or training) the computer how to perform these skills.
  2. Deploying the computer to actually perform these skills.

I won’t delve deeply into how we train computers to perform these skills – this usually falls into the domain of “creating or training AI Models,” and there are various elements that go into creating these models (insert all the terms we hear like deep learning, machine learning, neural networks, structure and unstructured data, etc.).

However, it is important to understand the distinction between AI and the approach we’ve pursued for the past century with traditional algorithms. When you programmatically code an algorithm, individuals provide direct instructions (the logic) to determine a specific outcome in response to a particular question (the input). In contrast, developing an AI model entails training it to navigate challenges for which it lacks specific instructions. This process can be thought of as cultivating “digital intuition.” It is this flexibility in an AI’s ability to adapt its logic that represents the most significant difference and the value proposition of AI compared to traditional algorithms.

For those who are interested in learning more about what goes into creating a model, I recommend reading IBM’s AI Overview or completing the “Artificial Intelligence Fundamentals” trail.

Is AI Dangerous?

Numerous science fiction movies have presented a bleak vision of the future, featuring some of the most frightening portrayals of AI. Consider films like “The Terminator” or “The Matrix,” where artificially intelligent entities have triumphed over humanity. I must admit, such a future is undeniably alarming. However, it’s essential to recognize that these scenarios depict what is known as Strong AI or general AI, which encompasses “generic” AI capable of executing any tasks based on human intuition and reasoning. Most experts argue that achieving general AI is not even feasible, so the fear of being overthrown by robots need not trouble us at present.

In reality, the majority of the contemporary AIs we interact with belong to the category of Weak AI, sometimes referred to as narrow AI. These are AI systems crafted to perform specific tasks, serving as highly advanced mathematical models in essence. While narrow AI may not offer the same dramatic appeal as Hollywood films, it presents a future in which efficiency and productivity can be significantly enhanced.

What Types of Tasks Can AI Perform?

Before an organization can define its AI strategy or even begin to explore the various AI-driven Salesforce features, it is helpful to categorize the various tasks AI can perform. Here’s how I personally like to think about the various categories:

  • Predictive – Traditionally, this refers to using AI to make numeric predictions (e.g., predicting lead conversion likelihood, opportunity closure likelihood, tomorrow’s temperature, Q3 revenue, etc.).
  • Generative – This category primarily focuses on using AI to generate content (e.g., text, images, audio) from scratch. This has seen significant growth thanks to the launch of AI models such as ChatGPT.
  • Categorize or Classify – AI in this category can be used to classify data (e.g., identifying a flower type from a picture, analyzing emails for phishing attacks, categorizing cases based on descriptions).
  • Robotic Navigation – This AI enables robots to perform various tasks and navigate environments that constantly change around them. Think of robotic vacuum cleaners that adapt to floor changes and shifting furniture.
  • Understanding Natural Language – AI in this category can extract the intention and meaning behind words. This is often referred to as Natural Language Processing (NLP), which allows models to interpret everyday language and respond meaningfully. NLP (like ChatGPT) plays a crucial role in advancing generative AI. Generative AI can refer to a computer’s ability to create a picture of a landscape, whereas NLP allows the computer to understand my request for a picture of a landscape.


AI can be described as the ability of a computer or machine to perform tasks typically executed by humans, requiring inference, intuition, and thought. We have never before had access to so many sophisticated AI models (e.g., ChatGPT, Einstein, IBM Watson, etc.), and the types of tasks AI can perform continue to evolve and expand daily.

In order for an organization to develop a strategy around AI, it’s important to be aware of the various applications of AI and the role that having good data has in getting value out of AI.

In my next post, I’ll discuss the evolution of AI in relation to Salesforce, explore the various areas of the platform where you can leverage AI, and highlight key considerations to take advantage of these features.




Andrew Cann

Andrew Cann is the VP of Delivery at Torrent Consulting. He is 13x Salesforce certified, is a member of the Salesforce Service Cloud Partner Advisory Board, and has worked in the ecosystem for 12 years.

Let's Work Together