Salesforce goes all in on generative AI with Slack GPT

With new applications of generative AI including Slack GPT and Einstein GPT, Salesforce expands ChatGPT-like AI automation capabilities across its CRM platforms.

At the Salesforce World Tour in New York on Thursday, the customer relations management giant announced a broad range of plans to inject generative, large language model artificial intelligence capabilities into its platforms, including Einstein, Slack and Sales Cloud.

The technology will enable AI-supported emails, qualified leads and sales planning. Salesforce’s chief marketing officer Sarah Franklin took the stage at New York’s Javits Center to explain the risks and benefits of AI in CRM, noting that the company will forge ahead with generative AI for sales and CRM, albeit with guardrails and “human in the loop” protocols.

Jump to:

Slack GPT is third LLM for the communication platform

Salesforce announced it is launching a natural language model AI into Slack, enabling low-code AI capabilities, various language models, as well as the power to add customer data insights from Customer 360 and Data Cloud.

Rob Seaman, who leads the enterprise product team for Slack, said there are now three LLM capabilities in Slack:

• Slack GPT, which comprises native features leveraging LLM within Slack for things like summarization, message drafting and composition of text.
• Integration of Einstein GPT, Salesforce’s LLM into Slack.
• An AI-ready platform designed with OpenAI, based on ChatGPT, that lets people build workflows that make calls out to any LLM of their choice.

The latter option comprises no-code workflows that embed AI actions with step-by-step prompts with plug-ins for other LLMs from such providers as OpenAI and Anthropic (Figure A).

Figure A

Seaman said the AI integration would allow end users and third parties to build workflows, apps and microservices around such functions as summarization and transcription.

Salesforce last week made the AI-ready platform designed with OpenAI available for developers of allied apps and services. Einstein GPT is available as a beta version, and the company is currently prototyping the integration of Einstein GPT within Slack, according to Seaman (Figure B).

Figure B

Usability and security are the goals

According to Seaman, key focuses in the development and deployment processes of infusing LLMs into Slack are end-user experience and security.

“A big part of what we want to do in Slack, one of our product principles, is don’t make the user think, so when these things are natively in Slack, we want to make it an experience that’s clear,” Seaman said. “Also, a lot of the most valuable information a company has, in the form of knowledge and communication, is in Slack.”

Seaman said the company is working on ways of leveraging the models on that data without that data actually traveling to and being stored in third-party data sources (Figure C). “Our enterprise customers would not be okay with that.”

Figure C

SEE: At the last NYC World Tour Stop Salesforce spotlighted Customer 360 with integrations for Slack, Tableau and Customer 360 (TechRepublic)

As part of this process, Salesforce developed no-data-storage and no-data-training agreements with LLM entities like OpenAI, so that if any data were to leave and go to any of the third-party LLMs it could not be stored or commingled with other data sources for the purposes of training.

As previously reported in TechRepublic, the company earlier this year also released guidelines to reduce bias in AI models around verifiability, safety, honesty, empowerment and sustainability.

Multi-prong LLM strategy and a large app ecosystem

Seaman said there have been some 4,000 third-party apps built on the Slack platform in the last two months that specifically work with LLMs, including a lot of apps that customers have built.

“You will hear us say there are 2,600 apps in the app directly, but there are many more than that built by our users themselves,” Seaman said.

The ChatGPT and Claude apps are examples with use cases around summarization and drafting, allowing questions and natural language prompts like “Hey, I’m trying to write an email to a customer about our newest product launch. Can you get me a template for a prospecting email to go to several high-tech customers?”

Einstein GPT

In March this year, Salesforce first threw down the ChatGPT- generative AI gauntlet by announcing it would add ChatGPT functionality to its Einstein CRM AI platform. Among other things, it includes:

• The ability to auto-generate prospect emails based on CRM data.
• A tool called Einstein Bots for Sales that automates lead conversion, pre-qualifying web visitors.
• Transcription and conversational AI analysis in multiple languages.

Workers want to use AI to cut non-productive work

The Slack State of Work report found 84% of senior IT leaders said generative AI will help their organization better serve customers and 67% of these leaders prioritized generative AI for their business within the next 18 months.

The report found desk workers who have adopted AI at their company are 90% more likely to report higher levels of productivity than those who have not. Those using automations at work reported saving an average of 3.6 hours a week.

At the World Tour, Salesforce also announced plans to collaborate with Accenture to accelerate the deployment of generative AI for CRM. Together, the companies intend to establish an acceleration hub for generative AI that provides organizations with the technology and experience they need to scale Einstein GPT helping to increase employee productivity and transform customer experiences, according to Salesforce.