My Thoughts On Salesforce Einstein Generative AI Features

Salesforce generative AI features will be generally available with Winter 24 release later this year (currently available in sandboxes) and this brings up an interesting question. Is your company ready for generative AI? Is it the right choice for you?

The company provides AI offerings for Service, Sales, Field Service and Commerce clouds. The last two are easier to adopt with a lower risk profile.

The Field Service one is a summary of the upcoming work for the day and this has a low risk of causing any issue in the workflow. The commerce offering is interesting, but I don’t think it’ll be used much as the product description is provided by a marketing team for a product and not something filled up at the last minute by the seller while creating a product entry. The only exception is an ebay-type website, but who would use Commerce Cloud to build that?

The sales and service offerings though deserves a consideration. The one I like the most is call summaries for sales teams which works as part of Einstein Conversation Insights. That said, it is not a new functionality as it has been around for a year or so now and this is what makes it ideal for demo and adoption.

The rest of the functionalities are around generative messaging for emails and chat in both sales and service context. This is one place where generative AI truly shines, but it is also one of those areas businesses might not be comfortable letting AI take over as of now.

The key concerns I see being raised are:

  1. Limited control on the tone and language used as this can have a legal impact.
  2. A lot of support centres have high-attrition rate and therefore business teams might not trust these users to make the right choice for an AI generated response.
  3. Response templates change often based on the fire that needs to be put out on any given day so AI generated templates might not be appropriate except for base support contexts. These too are better addressed with email templates, auto-response rules and macros.

To conclude, generative AI for internal consumption would be easier for business to adopt, but it would have low ROI. Customer facing generative AI is a double edged sword since even one mistake could be used to disable it and it would never considered again by a lot of companies.

What do you think? Leave a Reply