Salesforce debuts Generative AI at Dreamforce

Salesforce debuts Generative AI at Dreamforce

| Marketing Technology

Salesforce launched a new AI assistant at Dreamforce, as well as an enhanced data cloud. Will this help them retain their lead in Enterprise AI?

At Dreamforce, it was Salesforce's turn to launch an AI assistant. In many ways, CoPilot is similar to the products launched by their competitors, but it does introduce some additional use cases for Generative AI.

Salesforce have long been at the forefront of the AI trend. For years, they've been gradually embedding their Einstein AI capabilities across all aspects of their product portfolio. Indeed, activating the various Einstein features within Sales Cloud is an excellent way for midsize companies to get started with AI. There has been a sense though that some of their recent talk about AI is more about bandwagoning on the buzz around Generative AI buzz rather than about delivering effective product solutions. Dreamforce did a lot to dispel that notion.

The Generative AI hype has slowed in recent weeks, due to concerns about accuracy and security. Businesses have introduced strict guidelines around its use in order to guard against potential data leaks. Meanwhile, well publicised discussions about the risk of hallucinations have led to consumer mistrust.

Copilot

A wave of enterprise Generative AI tools have been launched to meet those concerns. Salesforce are no exception. The launch of a Siri-style Generative AI assistant called Einstein Copilot mirrors similar product launches from their competitors. Much comment was about the integration between Google's AI assistant and Salesforce's AI assistant. Much more interesting is the positioning and use cases envisioned for the new product.

Salesforce talked about the typical use cases for generative AI in an enterprise context - sales content creation, account summaries and meeting transcription. Developer support was a big area of focus for Copilot with the launch of the Copilot Studio. As with all Salesforce solutions, this supports admins in making both code and no code customisations.

Copilot Studio

A dedicated prompt builder tool is an interesting inclusion, which will help in crafting and refining AI prompts to get the best results. Expect more vendors to copy this approach as it's a good tool for training users, but it is particularly important for the sprawling Salesforce platform with its multiple customisation options. Even experienced Salesforce developers can struggle to keep up with the best ways to create a workflow for every use case.

The keynote added another rarely discussed use case for AI assistants: data discovery. Einstein Copilot was pitched as an extension of the Salesforce Data Cloud within the CRM. This ties into the other big product announcement from Dreamforce, Einstein 1, which is an AI layer on top of the existing Salesforce Data Cloud.

Einstein 1

The keynote extensively discussed metadata frameworks for integrated applications across Salesforce and third-party applications. This isn't actually a new concept. It's a redefinition of the existing approach used by many CDP vendors to combine datasets into a single customer view. Direct integrations with Snowflake and AWS data lakes make importing third party app data easier, but they're not a revolutionary new capability.

Data integration and aggregation have been a focus area for Salesforce for a while now. They were late entering the CDP space, and have been trying to catch up ever since. Including an easy to use AI layer is an easy differentiator for them in a crowded market. It allows for automated data integration and the development of custom AI models on top of the Salesforce platform. None of these capabilities are unique to Salesforce, but few enterprise technology companies do AI quite as well, particularly for firms looking for a no-code solution.

Perhaps the most significant change to Data Cloud was in the pricing announcement - Data Cloud will now be bundled with Enterprise editions of the CRM. This may encourage more companies to start using the Data Cloud capabilities and ultimately purchase additional licenses or storage capacity.

Banner Photo by Denys Nevozhai / Unsplash

Written by
Marketing Operations Consultant at CRMT Digital specialising in marketing technology architecture. Advisor on marketing effectiveness and martech optimisation.