By 
Who needs friends when you can converse with your data clean room, chat with your ad platform and, uh, talk to your marketing cloud?
On Wednesday, Salesforce announced a series of new features for its marketing and commerce clouds that use generative AI technology to help marketers be more efficient with their campaign creation, personalization, segmentation, measurement and offer customization.
Marketing GPT and Commerce GPT, both generally available, rely on Einstein GPT, the generative AI version of Einstein, a machine learning-based system that Salesforce introduced in 2016 for analyzing large CRM data sets across multiple sources.
With its newer AI investments, Salesforce is looking for inspiration from consumer-facing versions of generative AI, including ChatGPT and other platforms, said Michael Affronti, GM of Salesforce Commerce Cloud, during a press briefing on Monday.
“We’re really watching what’s happening in the consumer space,” he said, “and taking what we think are some of the best concepts.”
All the GPT
In March, Salesforce rolled out Einstein GPT through an integration with OpenAI.
“GPT,” which stands for “generative pre-trained transformers,” is a type of deep learning model first developed by OpenAI that gives applications the ability to create human-like text or images. Virtual assistants answer questions in a conversational manner – when they’re not hallucinating by responding with wrong or biased information.
The OpenAI relationship allows Salesforce to bring generative AI capabilities to all its various clouds and to other Salesforce-owned tools, including Slack for collaboration, Tableau for analytics and MuleSoft for API automation.
Marketing GPT
So, what’s new?
Through Marketing GPT, marketers can build segments, create scalable personalized content and automate journey creation.
For example, a marketer can use a natural language prompt to “interface” with Salesforce Data Cloud and create new audience segments for targeting, said Stephen Hammond, GM of Salesforce Marketing Cloud. Think of it as querying CRM data, but without having to know any code.
That marketer could then prompt the system to automatically generate personalized email versions and subject lines for testing, which the human is able to tweak, use or veto.
To help scale personalized content generation, Salesforce also announced a new partnership with Typeface, a recently launched AI startup that applies generative AI to content creation for enterprises.
On top of that, Salesforce is releasing a new tool it calls Segment Intelligence for Data Cloud, which automatically connects first-party data and revenue data with third-party paid media data from Google and Meta so marketers can get a better and more complete sense of campaign performance.
And Salesforce also rolled out an AI-powered identity resolution capability that uses AI to match consumer identities across its systems.
Commerce GPT
Keeping with the AI theme, Commerce GPT gives marketers similar tools, like the ability to prompt the system to quickly create multiple versions of content or to autogenerate product descriptions.
A marketer, for example, could type in “I want to liquidate an outdated product line,” and Commerce GPT will come back with recommendations for storefront design, a merchandising setup and promotions to run.
From there, the system will help the marketer target past buyers and set up promotions inside the Salesforce Marketing Cloud, Affronti said.
“We’re hearing from our customers and observing in the technology landscape that anywhere a customer wants to interface with your brand needs to be a transactable, elegant moment,” he said, in addition to being “a continuous experience, both in person and digitally.”
In that vein, Salesforce also released a chatbot tool as part of the new Commerce GPT that customers can use to start AI-generated conversations with customers across different channels, including messaging apps and digital storefronts.
Security questions
Although the first-party CRM data stored within Salesforce Data Cloud is what primarily powers Einstein GPT – Hammond described it as the “connective tissue” for all of Salesforce’s clouds – it isn’t the only place marketers keep their data.
“We recognize that there will be times when customers have data outside of our systems,” Hammond said.
To link data from outside of Salesforce with a customer’s CRM data, Salesforce announced an expanded relationship with Google Cloud Platform (GCP).
The partnership will let marketers bring their own custom AI models to Salesforce through Vertex AI, a machine learning platform within GCP that lets developers train and deploy ML models.
Through BigQuery, a data warehouse within GCP, marketers can use a method known as zero-copy integration to securely access data between systems without needing to copy the data or physically move it.
“None of the data that we have in our systems will be exchanged,” Hammond said. “The only time that information is ever even known to those external systems is at the time of query, and then it disappears – it’s a secure transaction.”
