Jul 23, 2018
How to natively integrate Cognigy.AI with Salesforce Einstein and the Salesforce Cloud for amazing customer experiences.
Salesforce’s outstanding solutions in marketing and sales technology power some of the most extraordinary customer experience solutions globally. With the acquisition of MetaMind in 2016, Salesforce added significant AI capabilities to its stack, ranging from machine learning based image recognition to advanced natural language understanding (NLU) for intent detection.
Integration with Salesforce’s generic Sales-, Service- and Marketing-Cloud APIs, as well as the newer Einstein APIs is something that many Cognigy customers regularly request, so in this article we will outline how such an integration is easily achievable, combining the power of Salesforce and Cognigy to deliver outstanding customer experiences.
Customer Service Solution
The logistics firm Pika TLS is looking for a way to handle a large portion of their currently manual customer service requests in an automated fashion.
For that, we're creating a customer service chatbot to handle the bulk of customer requests automatically. we’re connecting Cognigy to Salesforce Einstein for intent mapping and to Sales Cloud to retrieve shipment information.
Leveraging Einstein NLP for Intent Detection
Salesforce Einstein offers the capability to train NLU models via well documented APIs (https://metamind.readme.io/). The training of such models is described in a tutorial in the Einstein Intent Quick Start (https://metamind.readme.io/docs/intent-quick-start-custom-classifier).
After an intent model has been trained, the model can be queried by hitting the API endpoint https://api.einstein.ai/v2/language/intent with the authorization token in the header and the modelId and document (the sentence to classify) in the body of the request. Einstein will then deliver a list of detected intents with probabilities back, which can in turn be used within Cognigy.
The process of connecting a third party NLU engine to Cognigy is described in our documentation (https://docs.cognigy.com/docs/nlu-connectors).
You can choose to use both Cognigy NLU and Einstein NLU in conjunction, for example Einstein’s Intent Mapper and Cognigy’s slot extraction.
Connecting to the Salesforce Cloud
Based on the results of the intent detection, various queries can be executed against Salesforce Sales Cloud (and all other Salesforce clouds) to further facilitate the conversation.
To connect Cognigy.AI to Salesforce, Cognigy.AI's Salesforce Extension can be used. This module allows Cognigy.AI editors to write SOQL (Salesforce Object Query Language) queries directly from within Cognigy and work with the results. Cognigy has access to all of Salesforce's objects (accounts, tickets, opportunities, etc) and can read and manipulate those directly in Salesforce (given the correct access rights).
Result
Thanks to the power of Cognigy and Salesforce, customers can now converse with the PIKA bot in natural language and retrieve information instantly without having to engage a human service agent, freeing up time for human agents to handle more complex cases.
