Chapter 10: Conclusion

This chapter will conclude our time together learning about artificial intelligence (AI) on Salesforce. We will go through some of the main points we have learned and draw out some key lessons to take away for the future. Then, we will have a look at additional areas that you can continue to explore after you finish this book.

In this chapter, we're going to cover the following main topics:

• Using the power of built-in features
• Extending with declarative features
• Knowing when to go beyond declarative features
• Choosing where to go from here

After completing this chapter, you will understand what you have learned from working through this book and where to go to get a deeper understanding of the various areas discussed.

Using the power of built-in features

We started this book by looking at the built-in features in the core Salesforce clouds. No clouds are more used than Sales Cloud and Service Cloud, so it made sense to start our exploration there. What we saw was a range of pre-built features such as Einstein Lead Scoring, Einstein Forecasting, and Einstein Case Classification that use powerful and simple pre-built machine learning (ML) models to accomplish very specific tasks that help optimize sales or service processes.

In general, the lesson of the first few chapters is that there is a lot of value in the out-of-the-box features, and this can be realized quickly if these features happen to be a good fit for your business requirements and you don't have significant compliance or enterprise architecture constraints in terms of what you can implement. You can see a selection flow in the following diagram:

Figure 10.1 – Feature selection flow

Some features...

Extending with declarative features

The three declarative features we covered in Chapter 6, Declarative Customization Options (Einstein Next Best Action, Einstein Prediction Builder, and Einstein Discovery) can be used along with the standard Salesforce platform features such as flows and Lightning pages to create advanced custom solutions for your AI business requirements.

Einstein Next Best Action provides a framework for providing in-context recommendations for what to do next. A combination of a declarative strategy builder and the ability to extend the built-in functionality with Apex at various points of the flow means that you can find ways of bending this feature to do far-flung things if you feel so inclined.

For instance, you could decide to create recommendations linked to different screen flows, each representing a different layout for filling out the core record and replacing the normal page layout mechanism. That would be a bad idea, as you already have Lightning...

Knowing when to go beyond declarative features

Throughout this book, we have highlighted a range of scenarios that might mean you can't—or shouldn't—go forward with a standard or even a declarative solution. Whether for compliance or goodness of fit, or due to a need for detailed algorithmic control, a third-party or custom solution can in many cases be the right decision. Just because you have good features on a platform, this doesn't mean these features can solve every requirement well.

First, it is worth noting that Einstein Platform Services, while developed by Salesforce, is actually from an architectural point of view more or less equivalent to a third-party API solution just with a bit of additional tooling added to make your life easier. You should therefore evaluate these alongside other third-party APIs in most cases.

Third-party API solutions in general can add substantial value if they are a close fit to business requirements. They are...

Choosing where to go from here

Working through this book, you will no doubt have noted a number of different areas where you might want to deepen your knowledge. This book surveys a vast field of functionality, and unfortunately, we can't cover everything about everything. In the following sections, we will propose a few different ways you can extend the foundation you have gotten from this book. The following diagram shows some paths you can take:

Figure 10.2 – Where to go from here

First, we will look at ways to extend your knowledge of Salesforce AI features.

Salesforce AI features

We now list some of the most relevant resources to extend your knowledge of Salesforce AI features:

• For a solid deep dive into the Einstein platform from a Salesforce perspective, you can follow the Get Smart with Salesforce Einstein trail on Trailhead, which can be found at this link: https://trailhead.salesforce.com/en/content/learn/trails/get_smart_einstein...

Summary

Congratulations! You have made it to the end of the book. Sprinkle yourself with gold stars and get ready to implement some Salesforce AI solutions. On a more serious note, while you now have the foundation to understand the architecture and design of AI features on Salesforce, this is a fast-moving area and you should prepare to keep up to date—for instance, by looking at some of the additional resources we have suggested.

Overall, the key message of this book is that you as an architect have an impressive array of options at your disposal when creating solutions for AI requirements on Salesforce. However, the sheer number of powerful options you have available and the many detailed limitations and considerations that pertain to them means that your job gets harder, not easier.

You have to make the right engineering trade-offs between the different features at your disposal in order to satisfy both your business users and the various organizational constraints...

Questions

1. What are the main approaches to implementing the AI features discussed in this book?
2. What paths can you take to enhance your knowledge of architecting AI features on Salesforce?
3. How do you envisage the process of coming up with the right architecture for an AI feature?