Salesforce Rolls Out Customization Tool for Einstein Copilot Generative AI Assistant

Salesforce announced the public availability of the Einstein Copilot generative AI assistant just last week, and now the company is rolling out more AI products and developer programs at TrailblazerDX 2024, its developer conference. On March 6, the CRM giant rolled out:

• Einstein 1 Studio, which can be used to customize the Einstein AI assistant.
• A Slack Developer Program.
• Research on how tech leaders and IT workers perceive AI adoption.

Einstein 1 Studio lets organizations customize the Einstein Copilot AI

Einstein 1 Studio (Figure A) is a set of low-code AI tools for building, customizing and embedding the Copilot AI assistant into an organization’s custom Salesforce data sets.

“The Einstein 1 platform is how every business can build trusted AI apps of the future built on trusted data and on metadata,” said Clara Shih, chief executive officer of Salesforce AI, during a press briefing on March 4.

Einstein 1 Studio includes three main functions:

1. Copilot Builder: Lets developers create custom AI actions, instructing Copilot what action to take in the course of work. Copilot Builder is now in beta.
2. Prompt Builder: Lets admins and developers embed custom generative AI prompts in their workflows. Prompt Builder is now in general availability.
3. Model Builder: A platform to build or import generative AI models into Salesforce workflows. Model Builder can integrate a wide variety of popular models, including OpenAI and Google Cloud’s Vertex AI. It is now in general availability.

All three functions of Einstein 1 Studio are available globally in multiple languages. Global sovereignty options will become available on a rolling timeline “in the coming weeks and months,” said Shih.

Salesforce’s effort to use company data with AI safely

Salesforce’s plan is to empower developers to work with generative AI, mostly using prompts, and to customize the use of generative AI with private business data. Salesforce emphasizes that, depending on a customer’s needs and preferences, Einstein Copilot and the custom functions built in Einstein 1 will use only proprietary data, not feeding any information out into public models. The idea is to feed information that is already used by an organization in Salesforce, such as IoT devices or data lakes and warehouses stored with Snowflake or Databricks, into Einstein 1 Data Cloud for AI-powered CRM.

SEE: AI helps build a “data culture,” according to one of the digital transformation lessons from Salesforce World Tour Australia. (TechRepublic)

“When I talk to CTOs, they’re saying I want to give myself the flexibility to move with the future as this technology unfolds,” said Alice Steinglass, executive vice president and general manager of Salesforce Platform, during the press briefing. “And I think it’s part of a larger strategy of having multiple layers that fit together.”

She defines the multiple layers as:

1. A trust layer.
2. An open data strategy.
3. Large language models.
4. AI functionality embedded into the flow of work.

Slack Developer Program provides more options for building Slack apps

Salesforce announced a new developer portal for Slack. The portal is a space to test ideas, explore beta features, access tooling and find resources. The Slack Developer Program is open to developers globally starting March 6.

The Slack Developer Program offers three tools:

1. Sandboxes, up to 10 Salesforce Enterprise Grid instances at a time.
2. New custom functions in Bolt for Python or JavaScript.
3. Support for scripting with the Slack command line interface.

Tech leaders push AI, while some IT pros struggle with implementation

Salesforce showed research conducted by Vanson Bourne and Salesforce among 600 IT pros in Australia, France, Germany, the U.K. and the U.S. between December 2023 and January 2024.

Overall, the report found a disconnect between the needs of business leaders and IT regarding generative AI. Of the IT pros surveyed, 79% say business leaders are “increasingly pressuring them to implement AI.” Balancing speed, business value and security when implementing new tech is a struggle for 48% of those surveyed.

Meanwhile, generative AI is the top technology IT feels pressured to onboard quickly, and 88% of IT pros are unable to support all AI-related requests.

The top five challenges for IT when it comes to implementing AI are:

1. Lack of relevant skills.
2. Data security.
3. Data quality.
4. Slowing of other initiatives.
5. Increased cost.

Surveyed IT pros rank security as their top priority when it comes to new device implementation, while business leaders rank speed of implementation as their top priority.