Some Thoughts About Einstein Platform Services and Salesforce AI Strategy
Salesforce recently added new artificial intelligence(AI) services to its Einstein platform signaling strong progress in its AI ambitions. Under the Einstein Platform Services name, Salesforce added three new services focused on natural language understanding(NLU) and object recognition.
The new capabilities are powered by the MetaMind platform which Salesforce acquired last year. Among the big enterprise software incumbents, Salesforce stands up as the most aggressive acquired in terms of AI startups with half a dozen acquisitions in the last 18 months. Not surprisingly, some of the acquired capabilities have been slowly surfacing as part of the Einstein platform.
NLU plays a center role in the new Einstein Platform Services offering. Einstein Intent Service allows developers to train Einstein in intent-actions related to natural language sentences. Einstein links intents to specific actions on the Salesforce platform. Another addition to the new platform is the Einstein Sentiment Service which provides sentiment analysis in text data determining whether a user feels positive, negative or neutral. That type of capability can become a powerful addition to Salesforce business applications which already process large amounts of text data .
Einstein Object Detection is the third service in the Einstein Platform Services suite. The new API enables developers to train and use models that recognize objects in images. Object Detection capabilities can be a strong differentiator in some of Salesforce’s newest areas of focus such as ecommerce(Salesforce acquired ecommerce powerhouse DemandWare last year).
Observations About Einstein Platform Services
The release of Einstein Platform Services shades some light onto Salesforce’s AI strategy. Here are a few observations that I think are worth discussing about the new release:
1 — More Like WDC and MCS: Einstein Platform Services functionally compares with cognitive API platforms such as Watson Developer Cloud, Microsoft Cognitive Services or the AI APIs in PaaS stacks such as Google Cloud or AWS. From that perspective, Einstein Platform Services indicates that Salesforce plan to be a force to be reckon with in the cognitive services market.
2 — Not an AI Application Development Platform: The release of Einstein’s new cognitive services also indicates that Salesforce plans to focus on AI APIs instead of an open platform that can host AI models created in frameworks such as TensorFlow, Torch, Caffe2 and any other of the dozen deep learning stacks that have inundated the market in recent years. In that context, Einstein Platform Services is unlikely to compete with cloud machine learning(ML) platforms such as Google Cloud ML, Azure ML or AWS ML.
3 — Leading the AI-First SaaS Trend: The new Einstein Platform Services offering clearly indicates that Salesforce is reinventing itself as an AI-first SaaS. The sophisticated capabilities of the Einstein platform might help Salesforce to distance itself from its top competitors that, with the exception of WorkDay, haven’t put the same level of investment into AI technologies.
4 — A PaaS Follower, No Longer a Leader: Despite its impressive capabilities, the release of Einstein Platform Services is another examples of Salesforce trailing PaaS leaders such as AWS, Azure, Google Cloud, Bluemix and even Alibaba Cloud when comes to innovation in new technology markets. Mobile, big data, IOT, machine learning and now AI are some of the transformative technology trends in which Salesforce has constantly followed the PaaS leaders with undifferentiated technologies. Einstein Platform Services certainly looks similar to the AI APIs platforms from Microsoft, IBM, Google and Amazon.