Smart Metadata search using Einstein
Salesforce, the leading CRM is focusing a lot on making the platform intelligent, intuitive and hence the introduction of Einstein.
There are lot of cool features using Salesforce Einstein such as Lead Scoring, Email Analysis, Prediction Builder, etc. Most of these features are built-in within the platform but there is only a little we can tweak around!
Luckily, from a development standpoint, we are provided with a handful of APIs to build intelligent apps. The important aspect of the Einstein APIs is that Salesforce handles most of the heavy lifting, and gives us a probabilistic outcome. The APIs are what is called the 'Classification Algorithms', meaning it classifies the input and assigns labels to it.
The APIs are broadly classified into two types, the Vision and the Language related APIs. More information on these APIs can be found here.
In this article, lets look at a simple use case of using Einstein Intent API to perform some of the day-to-day org related activities.
Intent API, is a natural language processing model to classify the text we enter. The way this works is as follows:
- We create enough training data, with the sentences and the desired intent.
Example:
**Sentence:** Writing articles is my favourite past time.
**Label:**Writing
- We make a simple API call, and ask Einstein to train the data.
- We give a new sentence, and Einstein magically (by magic, we mean it uses the NLP and other complex stuff!) labels the sentence!
Example:
Sentence: Hey Einstein, i love to write!
Output label: Writing.
Now that we know the basics,lets get into action!
Below is the sample prototype of using Einstein in a lightning component. The result is a Smart Bot!