Watson & Einstein — AI Explained
“How do Watson & Einstein Work Together?”
That is the most popular question I’ve received over the last year+. Another popular question is “When do I use Watson and when do I use Einstein?” There is no one single answer as it all depends on your use case. Here I will explain and provide some examples of how the two AI platforms (IBM Watson and Salesforce Einstein) work together in specific scenarios, and which does what.
Conversational solution — Watson Assistant and Einstein Bots
If you are building a chatbot / virtual assistant, there are a few different ways to use Watson and Einstein together. In this first approach, Einstein Bots can handle your simple dialog flows for standard Q&A in English, provide responses to customers by accessing Salesforce customer data natively, such as profile or case information, and also make call-outs via Apex classes (you need to write and own the code) which is a handy feature. Watson Assistant can augment Einstein and handle your complex dialog flows, provide disambiguation when questions are not clear, allows digressions to provide answers but always keep conversations on track, multi-lingual support and also integrate natively and seamlessly with Watson Discovery (no code required) for answers buried in knowledge articles from multiple external data sources across your enterprise.
The benefit here is that you get to use both platforms where they can accelerate your business process together. The potential drawback is that you need to maintain your conversational solution across two different offerings so things like analytics will get tricky.
An alternative approach would be to use Watson Assistant tooling to handle all of your dialog, intents, entities, etc and pass through all conversation through the Einstein Bots platform. This would allow for a more seamless user experience to manage everything in one offering. Watson also requires significantly less intent examples to train on your intents (as few as 5 examples) whereas Einstein requires a minimum of 20 and suggests a minimum of 150 examples, so you can save additional time leveraging Watson’s natural language processing in this scenario as well. You can invoke Einstein Bots for specific things, as applicable for your use case, such as if you need to query Salesforce profile data.
Note: You can integrate IBM Watson into the Salesforce platform with as little as 5 lines of code using the Watson Salesforce SDK so its quick and easy!
Next Best Action With Watson & Einstein
Next Best Action With Watson for Salesforce Service Cloud is an IBM Watson AI-powered solution that augments and optimizes the agent experience to help them work more quickly and efficiently to solve their customer cases. Watson provides world-class intent classification and other key differentiating features such as out of the box tone analysis, custom domain entity extraction to understand your specific business and allows for automatic email draft responses. It provides agents with all the insights they need to take an action to resolve a case. You can see a demo of this in action below.
Watson also provides the AI-powered capability to create and run predictive models using the Watson Machine Learning service. With this, you can make predictions on various things such as the likelihood for a case to be escalated or the likelihood for a lead to successfully close.
Einstein Next Best Action is a solution that provides recommendations processed by action strategies defined by business rules. These are detailed workflows to guide agents through executing a specific action for a case. As an admin, you build out the steps of the flow, and strategy, for each of your workflows that would guide agents.
Here, Watson and Einstein can work together where Watson provides the AI-powered insights and predictions required for Einstein’s strategy engine to identify recommendations on what action to take and data required to complete it. Einstein can then guide the agent through a step-by-step workflow to execute on that recommendation.
Watson Discovery For Salesforce & Einstein
Watson Discovery for Salesforce is applicable to almost any Einstein use case/technology you have implemented, or are planning to implement, in Service Cloud. With this differentiating capability, Watson augments the Einstein experience, whether its a virtual agent using Bots, Next Best Action for recommendations or other Einstein features. Watson leverages its powerful AI capabilities to provide agents with answers and information they need that are buried in documents such as knowledge articles, employee manuals and more, from across your entire enterprise. It also helps agents solve their cases more quickly by providing relevant, previously resolved Salesforce cases to review. The Watson Discovery for Salesforce app is available for free on the Salesforce AppExchange and with a free Watson Discovery service on IBM Cloud, you can deploy this app in as little as an hour to see immediate value.
The Bottom Line — Watson & Einstein
There are many ways that you can leverage Watson & Einstein together to achieve your goals. I provided a few examples above to get you started but there are alot more details, possibilities and many combinations of solutions across the IBM Watson platform and the Salesforce Einstein platform. The key to start is to clearly define your use case and requirements, then understand what both Watson and Einstein can do, and after that, create a solution that best meets your business needs.
Marc is the CTO for IBM Watson AI Strategic Partnerships. When not leading the technology vision and strategy for IBM Partnerships, Marc enjoys DJing, playing video games and wrestling.
