Salesforce Einstein and the Emergence of the AI-First SaaS
Dreamforce is around the corner and Salesforce is once again set to unveil some major technology releases. While topics such as mobile, IOT and analytics have dominated previous editions of Dreamforce, this year the focus seems to be 100% around artificial intelligence(AI) and the newest member of the Salesforce family: Einstein.
Based on initial press coverage, Einstein will enable AI capabilities in a large number of products of the Salesforce platform. The introduction of Einstein should be seen as a new phase in the Salesforce platform using AI as a first class citizen. From the perspective of the SaaS ecosystem in general: the release of Einstein could mark the beginning of the AI-First SaaS.
Five Capabilities of AI-First SaaS
AI offers a new canvas to reimagine many of the existing SaaS capabilities and create new ones using AI as the foundational element. There are many interesting features that could be relevant to AI-First SaaS platform. Here are some ideas that I believe are worth considering.
Prebuilt Machine Learning Models for Well-Known Business Processes
An AI-First SaaS should include predefined machine learning(ML) models that improve existing business processes. In the case of Einstein, it is conceivable that the platform will include ML models that operate against Salesforce marketing or sales data to enable new capabilities such as predictive sales forecast, funnel optimization, personalized product offering among many other relevant capabilities of sales and marketing processes.
Building and Running Custom ML Models
In addition to executing predefined ML models, an AI-First should provide the capability of building, running and managing custom ML models. In the case of Salesforce Einstein, it would be interesting if the platform supports the execution of custom R or Python models against Salesforce data extending the default ML feature set.
Model Training and Monitoring
In order to support mission-critical business processes, an AI-First platform should enable the continuous performance monitoring and training of ML models. In the case of Salesforce Einstein, the platform could include tools within the Salesforce management console to retrain the existing models, monitoring its runtime behavior and evaluate its effectiveness.
Deep Learning Capabilities
An AI-First SaaS should take advantage of the recent developments in deep learning technologies to enable the analysis of cognitive and unstructured data sources such as text, vision and speech. In the case of Salesforce Einstein, we can envision the platform enabling new analytic capabilities using image, audio or text which are highly relevant in sales and marketing processes.
An AI-First SaaS should expose APIs that allow third party applications to leverage AI and ML models. In the case of Salesforce Einstein, its conceivable that the new AI capabilities will exposed a new set of APIs for programmatic interactions with other applications.
These are just some intriguing ideas for Salesforce Einstein. I am definitely looking forward to Dreamforce to see Salesforce’s strategy to lead the AI-First market.