Posted on August 21, 2023 in Artificial Intelligence 

The Definitive Guide to Einstein GPT

Announced at TrailblazerDX 2023, Salesforce is launching Einstein GPT to deliver auto-generated AI content, right within the Salesforce platform. There are many use cases for Einstein GPT, including Slack, sales, service, marketing, commerce, and app builders. This is the world’s most powerful, intelligent, generative CRM – making every employee more productive and every customer experience better.

In this article, I’ll go over everything you need to know about Einstein GPT.

What is Einstein GPT?

Einstein GPT is a powerful AI tool that combines public and private AI models with CRM data. This allows users to ask natural-language prompts directly within Salesforce CRM. The result is AI-generated content that is continuously adapted to changing customer information and needs. This can save users a significant amount of time, as they no longer have to manually create content.

What do we mean by public and private AI models? Einstein GPT infuses:

• Salesforce’s private AI models: In other words, those developed by Salesforce, which is leveraged by customers using features from the Einstein technology layer.
• Combining OpenAI’s ChatGPT: Thanks to the Salesforce and OpenAI partnership, providing out-of-the-box generative AI capabilities to companies through direct access to OpenAI’s enterprise-grade ChatGPT technology.
• Bring your own model: As Einstein GPT is open and extensible, you can opt to use it alongside your own external model/s.

Data is being pumped into these models, in real-time, from Salesforce ‘clouds’, and the Salesforce Data Cloud to ingest, harmonize, and unify all of a company’s customer data.

Einstein GPT Trust Layer

The Einstein GPT Trust Layer is a new industry standard for trusted enterprise AI that is part of AI Cloud. It allows organizations to benefit from generative AI while also being reassured about data privacy and security. This is because the Einstein GPT Trust Layer prevents large-language models (LLMs) from retaining sensitive customer data.

Multiple Salesforce executives have expressed that trust and privacy were the primary starting point when designing their generative AI capabilities – a standpoint they believe differentiates them from most of the market. 

Salesforce has done work to adhere to data privacy and confidentiality. One example is product description generation in Commerce GPT. The process happens at the time of the query (prompt), where the most limited amount of data gets passed over the wall to the LLMs that power the generative capabilities. This is self-destructive, and as a result, customer data never leaves the Salesforce cloud products where it’s stored. 

This separation of sensitive data from the LLM will help customers maintain data governance controls while still leveraging the immense potential of generative AI.  

Anecdotes from Salesforce executives follow the same thread – that Salesforce customers are more willing to pilot generative-AI technology when they become aware of the ‘safety net’ that Salesforce has incorporated (as opposed to remaining hesitant if it were an open API call out to other providers). 

The following diagram shows how the Einstein GPT Trust Layer works:

As depicted in the diagram above, the Trust Layer of Einstein GPT addresses the apprehensions regarding risks that come with adopting generative AI. It accomplishes this by allowing customers to fulfill their enterprise data security and compliance requirements, while still reaping the advantages of generative AI.

There are some phrases in the above diagram that you may be unfamiliar with, so below I’ve broken some of them down:

• Prompt: A piece of text that is used to guide the generation of text. It can be as simple as a single word or phrase, or it can be a more complex set of instructions.
• Dynamic Grounding: A technique in natural language processing that allows a model to understand the context of a word or phrase based on the surrounding text. This is in contrast to static grounding, which relies on a pre-defined dictionary or thesaurus to map words to their meanings.
• Data Masking: Data masking is a process of replacing sensitive data with non-sensitive data, also known as placeholders, to protect the confidentiality of the data. 
• Toxicity: In the context of generative AI, toxicity refers to the production of harmful or offensive content.

Einstein Studio

Einstein Studio offers an innovative solution for businesses to leverage AI effectively while maintaining data privacy. By seamlessly integrating your organization’s data with external AI models, Einstein Studio provides a powerful way to harness AI’s capabilities without compromising data ownership.

This option allows you to tap into the strength of various predictive and generative AI services while ensuring your data remains secure within your organization. And by combining the power of Einstein Studio with the security and compliance features of the Einstein GPT Trust Layer, businesses can build and deploy generative AI models with confidence, knowing that their data is protected and their models are compliant with regulations.

With the Bring Your Own Model (BYOM) approach, data science and engineering teams can effortlessly create, train, and deploy customized AI models within Salesforce. This empowers you to enhance Salesforce applications with tailor-made AI predictions and auto-generated content, streamlining workflows and decision-making processes. The BYOM method eliminates the complexities associated with AI integrations and model training, saving time and resources. Moreover, the framework operates on a zero-ETL principle, seamlessly incorporating advanced AI models with relevant data across various departments such as sales, service, marketing, commerce, and IT. This ensures optimal utilization of existing data without the need for intricate integration efforts.

Einstein Studio offers a user-friendly interface designed to facilitate the integration of your customized external AI models with your CRM data. The platform guides users through every stage of this distinctive integration process.

The Model Builder component within Einstein Studio assists in selecting the appropriate model type. Users have the flexibility to:

• Build a new model from the ground up.
• Connect an existing SageMaker model.
• Integrate a model from Google Vertex AI.
• Link a model located on another external platform.

Einstein Studio emerges as a solution to a significant AI adoption challenge in the tech industry. It simplifies the process of merging company data with preferred AI models from other predictive or generative AI services. This empowers users to effectively translate AI insights into actionable outcomes.

In the realm of AI models, Salesforce introduces Einstein Studio, which not only offers their proprietary LLM but also accommodates popular options like Amazon SageMaker and Google Vertex AI. This underscores Salesforce’s commitment to an open ecosystem approach, enabling organizations to select the preferred providers for their teams.

Prompt Studio

Salesforce Prompt Studio is a new feature in Salesforce that allows you to create prompts that can be used to generate personalized content. Prompts are short, natural language questions that you can ask Salesforce to generate content for. For example, you could ask Prompt Studio to generate a personalized email, a social media post, or a website page.

Prompt Studio is powered by a large language model (LLM), which is a type of artificial intelligence (AI) model that is trained on a massive dataset of text. This allows Prompt Studio to understand the meaning of natural language and to generate content that is both informative and engaging.

Prompt Studio is based around templated solutions for your users. So instead of giving them an interface such as ChatGPT where they can ask any question, and most likely have to input all of the information. Templated solutions can be created for specific scenarios such as “Write an introduction email to this customer” or “Write a follow up email to this case”. 

To use Prompt Studio, you simply need to type a natural language question. For example, you could type “generate a personalized email to my customer Chris Smith.” Prompt Studio will then generate a personalized email that is tailored to Chris Smith’s specific needs and interests.

Here are some benefits of using Salesforce Prompt Studio:

• Personalized content: Prompt Studio can be used to generate personalized content that is tailored to specific customers or audiences. This can help to improve customer engagement and drive sales. For example, you could use Prompt Studio to generate a personalized email for a customer who has just signed up for your service. The email could include information about the customer’s specific interests and how your service can benefit them.
• Engaging content: Prompt Studio can be used to generate content that is relevant and interesting to customers. This can help to keep customers engaged and coming back for more. For example, you could use Prompt Studio to generate a social media post that is about a topic that your customers are interested in. The post could include a link to your website or blog, as well as a call to action.
• Sales-driving content: Prompt Studio can be used to generate content that is persuasive and that encourages customers to buy. This can help to increase sales and grow businesses. For example, you could use Prompt Studio to generate a sales pitch that is tailored to a specific customer’s needs. The pitch could include information about the customer’s specific pain points and how your product or service can solve them.

Overall, Salesforce Prompt Studio is a powerful tool that can be used to improve customer engagement, drive sales, and grow businesses. However, it is important to use the tool in a way that is effective and ethical. For example, you should make sure that the content that you generate is accurate and error-free. You should also avoid using the tool to generate content that is discriminatory or offensive.

GPT Suite of Products

Sales GPT

Salesforce Sales GPT is designed to help sales teams create more personalized and engaging content. It can be used to generate content that is tailored to specific customers or audiences, and it can also be used to create content that is relevant to the latest trends and news.

Sales GPT offers the following features:

• Sales Assistant: The tool summarizes every step of the sales cycle in a side panel, from account research and meeting preparation to drafting contract clauses. It also automatically keeps the CRM up-to-date, so you can be sure that your sales data is always accurate and up-to-date.
• Sales Emails: The tool can automatically generate personalized emails for every customer interaction, using data from your CRM system. This means that you can send emails that are tailored to the specific needs and interests of your customers.
• Call Summaries: The tool automatically transcribes and summarizes calls, and then sets follow-up actions based on the transcription. This can help to improve seller productivity by freeing up time that would otherwise be spent manually transcribing calls and setting follow-up actions.

Service GPT

Salesforce Service GPT is a generative AI tool that can be used to improve customer service and field service operations. It can help to improve customer satisfaction, reduce costs, increase productivity, and make better decisions.

Service GPT offers the following features:

• Service Replies: The tool can automatically generate personalized responses based on real-time data sources, such as your CRM data and other sources. This means that you can send responses that are tailored to the specific needs and interests of your customers.
• Work Summaries: Create concise and informative summaries of service cases and customer engagements based on the details of the case and the customer’s history.
• Call Summaries: The tool automatically transcribes and summarizes calls, and then sets follow-up actions based on the transcription. This can help to improve seller productivity by freeing up time that would otherwise be spent manually transcribing calls and setting follow-up actions.
• Knowledge Articles: The tool can automatically generate and update articles based on the latest real-time data from support interactions. This means that the articles are always up-to-date with the latest information and best practices.
• Mobile Work Briefings: The tool summarizes critical information about each appointment before field service teams arrive, helping them to work more efficiently. This information includes the customer’s contact information, the issue that needs to be resolved, and any relevant history.
• Engagement: Salesforce Service GPT can be used to generate content that is relevant to the latest trends and news. This can help to keep customers engaged and coming back for more.

Marketing GPT

With Marketing GPT, Salesforce aims to give marketers an AI-connected user interface that augments campaign brief conception, audience/segment discovery, and content creation. 

Marketing GPT offers the following features:

• Segment Creation: Marketers are given the ability to create audience segments quickly and improve targeting using natural language prompts to query Data Cloud, Salesforce’s CDP offering.
• Email Content Creation: This gives marketers the ability to build email body content, create subject lines, and more, automagically.
• Segment Intelligence: Segment Intelligence steps in to help marketers understand campaign performance, relative to an audience segment. 

Commerce GPT

Commerce GPT is a generative AI tool that can be used to provide personalized commerce experiences throughout the entire buyer’s journey. 

It can be used to generate auto-generated recommendations, content, and communications that are based on real-time data from the Data Cloud. This ensures that the recommendations and content are relevant and timely.

Commerce GPT offers the following features:

• Goals-Based Commerce: This tool empowers businesses to set targets and goals, and then provides actionable insights and proactive recommendations on how to meet them – for example, to improve margins or to increase average order value (AOV).
• Dynamic Product Descriptions: With Commerce GPT alleviating the work of creating product descriptions, new storefronts are faster than ever to ‘stand up’. It can also automatically fill in missing catalog data for merchants and revolutionize the customer experience with auto-generated product descriptions tailored to every buyer.
• Commerce Concierge: This application experience built on top of Commerce GPT uses bot technology and generative technology to help brands deliver a 1:1 shopper experience on any messaging channel. 

Commerce GPT is a powerful tool that can help brands create more engaging and personalized shopping experiences for their customers. It is a key part of Salesforce’s vision for the future of commerce, which is to provide customers with the ability to shop seamlessly across channels.

Slack GPT

Slack is a cloud-based communication platform that helps teams to collaborate more effectively. It provides a centralized space where teams can communicate in real time, share files, and collaborate on projects. Slack can also be integrated with other tools and services, which allows teams to streamline their workflows and work more efficiently.

Customers rely on Slack to store their institutional knowledge in channels dedicated to each project, team, or topic. They also securely integrate their tools in Slack using our open, extensible platform and partner ecosystem. Slack GPT is being built on this same foundation, bringing trusted generative AI to where your team already works: in Slack.

Slack GPT offers the following features:

• An AI-ready platform to integrate and automate with your language model of choice, whether you use partner-built apps such as OpenAI’s ChatGPT, Anthropic’s Claude or build your own custom integration.
• A set of AI features built directly in Slack, including AI-powered conversation summaries and writing assistance
• A new Einstein GPT app that lets you surface AI-powered customer insights from trusted Salesforce Customer 360 data and Data Cloud.

Tableau GPT

Tableau is a data visualization and business intelligence software that allows users to connect to various data sources, create interactive dashboards, and generate insightful reports. It helps users analyze, understand, and communicate data in a visual and interactive format.

Tableau is a powerful tool that can be used by businesses of all sizes. It is easy to use and does not require any coding knowledge. Tableau can be used to connect to a wide range of data sources, including databases, spreadsheets, and cloud services.

Tableau GPT is a new feature that is being added to Tableau. It uses generative AI to automate many of the tasks involved in data analysis. This makes it easier for users to analyze data and create visualizations.

Tableau GPT is still under development, but it has the potential to revolutionize the way data is analyzed. It could make data analysis more accessible to a wider range of users and help businesses make better decisions based on data.

Apex GPT

Apex GPT is a generative AI tool that helps developers and IT teams to improve software development speed and efficiency. It can be used to generate intelligent Apex or LWC code through natural language prompts, which can help to speed up the development process.

Apex GPT also has the ability to scan for code vulnerabilities and provide inline code suggestions directly from the Salesforce integrated development environment (IDE). This can help to improve the quality of code and reduce the risk of errors.

Overall, Apex GPT is a powerful tool that can help developers to improve their productivity and the quality of their code.

Flow GPT

Flow GPT is a generative AI tool that allows users to create workflows based on a single text prompt. This means that users can simply type a natural language description of what they want the workflow to do, and Flow GPT will generate the code for the workflow.

For example, a user could type “I want to create a workflow that sends a notification to sales representatives when a lead is converted to an opportunity.” Flow GPT would then generate the code for a workflow that does exactly that.

Flow GPT is a powerful tool that can help users to automate their workflows and save time. It is also a great way for users who are not familiar with programming to create workflows.

Pricing

The pricing of Sales GPT & Service GPT were announced a few weeks ago, and will be rolled into the existing Sales & Service Cloud Einstein Licenses, priced at $50/user per month. However, if you have Unlimited Edition licenses, Sales & Service GPT will be included. 

This includes a limited number of Einstein GPT credits (number TBC), which go towards generating outputs from user prompts. Currently, it’s not confirmed what constitutes one credit. Other large language models (LLMs) work on the basis of tokens, which determine the scope of information that the model can analyze when generating outputs. 

Salesforce is releasing Enterprise Expansion Packs to democratize pricing for generative AI. Currently, Salesforce has to absorb the cost of processing data when using their LLM (i.e. Einstein GPT), which can be expensive. To make pricing more fair, Salesforce is offering a pay-as-you-go option for customers who exceed their allotted credits.

Enterprise Expansion Packs for Sales and Service teams provide additional credits for querying large databases. This means that teams can use more generative AI without having to worry about exceeding their allotted credits.

Trailhead Badges

Earlier this year, Salesforce announced that it would be introducing 35 new AI-focused Trailhead badges. Some of these have been released, with others soon to follow.

Not all 35 badges have been released yet, but there are still plenty of badges to help you get started in learning everything about Einstein GPT. From AI and Generative AI fundamentals to quick look badges of every GPT product, there’s plenty to get your teeth into. Check out our guide to the top AI badges below. 

Summary

It’s safe to say that Einstein GPT has some extraordinary expectations to live up to, but Salesforce have already demonstrated how this AI technology endeavors to be safe and secure without limiting its capabilities.

It’ll be exciting to see its use cases once we begin using it in full force.

Read more:

• Breaking News: Salesforce Announces AI Cloud for Einstein GPT
• Salesforce Announces Einstein GPT Integration with Flow + Data Cloud
• Available Now: Parts of Sales GPT, Service GPT, and Einstein Trust Layer
• Salesforce Reveals Einstein GPT Pricing
• A First Look at Salesforce Prompt Studio: Every Admin is a Prompt Engineer