What Is Agentforce Assistant? (Formerly Einstein Copilot)

Editor’s note: Agentforce Assistant has been upgraded to Agentforce. This article was originally published on March 6, 2024, and was updated on May 23, 2024.

Salesforce continues to expand the capabilities of Agentforce Assistant, its customizable, conversational, generative AI assistant. Here’s the latest on what it is, what organizations can do with it, and how it will continue to evolve to meet critical business needs.

Agentforce Assistant is natively embedded across Salesforce applications to deliver a consistent user experience that can answer questions, generate content, and dynamically automate any action to deliver better productivity, deeper customer relationships, and higher margins.

Unlike other AI assistants or copilots that lack adequate company data to generate useful responses, Agentforce Assistant uses an organization’s own unique data and metadata to produce powerful customer insights and recommendations, while maintaining privacy and data governance and without requiring costly AI model training.

For example, using simple natural language prompts, sales professionals can employ Agentforce Assistant to query past customer transcripts to assess sentiment, and craft followup emails based on previous calls. Marketers, meantime can use Agentforce Assistant to build more intelligent campaigns that fit all of their business’s marketing needs in one simple brief or use customer and brand data to create personalized campaigns that really resonate with their target audiences. And merchants can use Agentforce Assistant to help optimize sales, build ecommerce websites, and quickly generate personalized promotions.

Agentforce Assistant comes with a library of pre-programmed capabilities, automated responses, and business tasks that the AI can perform for users when prompted. Actions can be combined to execute dynamic multi-step plans.

What you'll learn:

• How do Agentforce Assistant, Data Cloud, and Salesforce Platform fit together?
• How does Agentforce Assistant empower teams to work faster?
• How is Salesforce extending Agentforce Assistant’s capabilities for different industries and use cases?
• What platform enhancements have been made to Agentforce Assistant?
• How does Salesforce ensure trust in Agentforce Assistant?
• How will Agentforce Assistant shape the future of CRM?
• How can customers get Agentforce Assistant?
• Go Deeper

Agentforce Assistant is part of Salesforce’s Salesforce Platform, which integrates the user interface, a variety of AI models, and data in a single metadata-driven platform. A trusted AI platform for customer companies, Salesforce Platform gives companies the ability to safely connect any data to build AI-powered apps with low code and deliver entirely new CRM experiences.

AI is only as good as the data it’s trained upon, and that’s why grounding Agentforce Assistant in Data Cloud makes it so powerful. Data Cloud connects, federates, and harmonizes any data type from any product and system, and connects it back to the Salesforce applications that business users need to use every day, to deliver a comprehensive, 360-degree view of customers and power CRM, AI, automation, and analytics across any business process.

BYOL (Bring Your Own Lake) means companies can integrate any system with Data Cloud, accessing all of their trapped data from data lakes like AWS Redshift, Google BigQuery, Databricks, and Snowflake within Salesforce. They also no longer need to move data across platforms because of Data Cloud’s zero copy or zero ETL (extract-transform-load), and it connects to external predictive models for seamless use in Salesforce workflows.

Also, for the approximately 80% of data that is unstructured, semantic search and vector embeddings in Data Cloud unlock a brand new way to interact with data. Agentforce Assistant Search enhances Agentforce Assistant, providing sales, customer service, marketing, commerce, and IT teams with an AI assistant capable of solving problems and generating content by accessing real-time unstructured and structured business data.

With Data Cloud, all the relevant, trusted customer and business data — structured and unstructured, within and beyond Salesforce, across many disparate data lakes — can be unified and made available for use by every team and workflow. With zero copy, it’s possible to share data among two or more data stores without actually moving it.

Agentforce Assistant brings together an intuitive interface for interacting with AI, world-class AI models, and deep integration of the data and metadata needed to benefit from AI.

In contrast to other AI assistants, it is a single, conversational copilot that grounds its responses in trusted and harmonized data from Data Cloud to provide the necessary context for high-quality outputs. What’s more, because Agentforce Assistant is natively integrated with all Salesforce CRM applications, customers can access insights and recommendations from it whether they’re in Sales Cloud, Commerce Cloud, Marketing Cloud, Slack, Tableau, or MuleSoft.

For example:

• Sales professionals can accelerate deal closures by summarizing records or generating customized communications to provide more personalized client engagement.
• Service teams can streamline case resolution and boost customer satisfaction by surfacing relevant answers, offers, and data from disparate systems.
• Marketing professionals can simplify the development of new campaigns, create website landing pages based on personalized consumer browsing and buying preferences, and easily create contact forms that populate based on unified customer profile data from across the Salesforce Platform. Following online purchases, Agentforce Assistant can also automatically generate customer surveys, which help businesses increase long-term engagement and encourage future orders.
• Ecommerce retailers can tap Agentforce Assistant for step-by-step assistance in building high-converting digital storefronts as well as customize and design different storefront components simply through conversation. Users can have Agentforce Assistant automate complex tasks like managing multi-product catalog data and with the power of generative AI, Agentforce Assistant can develop product descriptions in multiple languages, personalized product promotions, and SEO metadata that help drive conversions.
• Financial service companies can use Agentforce Assistant to trigger actions like triggering fee reversal requests for disputed transactions, issuing provisional credits, and pulling lists of recent customer transactions – saving time and resources.
• Developers can supercharge their productivity by prompting Agentforce Assistant using natural language to generate Apex code, suggest more effective and accurate code, and proactively scan for any code vulnerabilities — all from within the developer environment.
• Data analysts working in Tableau can enjoy productivity gains. Using the conversational interface, analysts can go from raw to actionable insights with just a few prompts. Agentforce Assistant helps perform faster data exploration, build relevant visualizations based on best practices, automate repetitive tasks, and promote efficient data curation.

Salesforce does this through actions. A copilot includes a library of actions, which are pre-programmed capabilities that enable Agentforce Assistant to not only answer questions using business data, but also string together workflows to get things done on behalf of users. For example, if a user asks an assistant for help with writing an email, the copilot launches an action that drafts personalized emails, grounded in relevant Salesforce data. Salesforce provides some standard, out-of-the-box actions but also lets users customize their own actions to meet the needs of their specific organization. New Agentforce Assistant capabilities, such as those available through Einstein 1 Studio, allow businesses to build and customize their own AI assistants with skills and prompts that fit their needs.

In addition, Salesforce is increasing its library of persona- and industry-specific actions that expand the range of pre-programmed capabilities Agentforce Assistant can perform. For example, the company has announced:

• Marketing actions: Agentforce Assistant for Marketing streamlines the campaign-creation process by automatically generating marketing briefs, content, and email campaigns. It also helps marketers speed up time-to-market with auto-generated responses, aligned to brand voice.
• Merchant actions: Agentforce Assistant for Merchants simplifies commerce tasks, like setting up storefronts, sending personalized promotions that delight customers, and taking steps to improve conversion rates.
• Sales actions: Agentforce Assistant Sales Actions boosts seller productivity across cross functional use cases, such as closing plans, forecasting guidance, querying prior call  transcripts captured in Einstein Conversation Insights to track customer sentiment, and send follow-up emails. With Agentforce Assistant in Sales Cloud Everywhere, customers can also take Agentforce Assistant everywhere they work.
• Analytics actions: Agentforce Assistant for Tableau helps users in every role and function dive deep into their data, using Tableau’s powerful analytical engine through natural language to query and derive rich insights from data sources like spreadsheets, cloud and on-premises data warehouses, and Salesforce Data Cloud.
• Education actions: Agentforce Assistant Recruitment & Admissions Actions helps educators and staff quickly answer prospective student questions with precise responses that are grounded in previous student interactions, knowledge articles, events, and other institutional knowledge.
• Banking actions: Agentforce Assistant Banking Actions automates time-consuming tasks like capturing and summarizing customer details, updating customer information, and personalizing email communications.
• Health actions: With Agentforce Assistant Health Actions, qualified healthcare professionals can use natural language prompts to capture and summarize patient or partner details from different clinical and non-clinical sources, easily update patient and member information, and automate outreach.

In spring 2024, Salesforce announced several platform enhancements to make Agentforce Assistant’s conversational AI experience even better, including:

Copilot Analytics: Admins can now visualize usage of Agentforce Assistant through a pre-configured analytics dashboard, enabling them to dissect key metrics such as actions used, average interactions per user, and success rates to help businesses drive and measure ROI from AI.
Recommended Actions: Teams can now complete any task with recommended actions in the flow of work. Instead of typing in a request, Agentforce Assistant surfaces single-click standard actions specific to the page a user is on, such as “summarize opportunity” on the opportunity page or “draft an email” on the contact page.
Agentforce Assistant on the Salesforce Mobile App: Agentforce Assistant is now live on the Salesforce mobile app, with voice-to-text functionality for mobile users that enables them to simply talk with their data and workflows.

Users can customize Agentforce Assistant and seamlessly embed AI prompts and actions across any CRM app with Einstein 1 Studio, a set of low-code tools deeply integrated with Data Cloud.

• Prompt Builder: A low-code prompt management tool that helps users create custom, reusable AI prompts in the flow of work, broadening the use of generative AI beyond conversational interfaces.
• Copilot Builder: A library of actions, or capabilities, that users can choose from to customize their AI assistant. For example, users can use it to configure their AI assistant to perform tasks like updating an account record or getting a shipping update.
• Model Builder: A capability that allows users to ‘bring your own LLM’ — including those from Anthropic, Hugging Face, Google, and OpenAI — to power custom Agentforce Assistant functionality.

Trust couldn’t be more important to today’s businesses. In fact, Salesforce research found the likelihood of business buyers using AI to improve their experiences declined in 2023 compared to the previous year. Another study found that most people (57%) do not trust AI.

Agentforce Assistant addresses such concerns by serving up trusted AI interactions with privacy and security measures provided by the Einstein Trust Layer. Every AI interaction passes through the Einstein Trust Layer, part of Salesforce’s Salesforce Platform, which can perform functions like masking personally identifiable information (PII), scoring outputs for toxicity, and helping to protect information from unauthorized access and data breaches through zero-data retention from Salesforce’s LLM partners.

New to the Einstein Trust Layer is customer-configured data masking, enabling admins to select the fields they want to mask, providing greater control. Additionally, the audit trail and feedback data collected from AI prompts and responses is now stored in Data Cloud, where it can be easily reported on or used for automated alerts through Flow and other Salesforce Platform tools.

Salesforce believes trusted AI needs a human at the helm. Instead of asking humans to intervene in every individual AI interaction, Salesforce is designing more powerful, system-wide controls that let humans focus on the high-risk items that really need their attention. In other words, humans let the AI row the boat but remain very much in charge of steering the ship. This means customers can get the very best from both human and machine intelligence.

Agentforce Assistant is the first glimpse into a future where people primarily interact with software using conversational interfaces rather than clicking buttons, and will fundamentally change the way people use technology. Going forward, using AI assistants will not just be about content generation and summarization. Agentforce Assistant will be able to proactively take action for users. For example, Agentforce Assistant will be able to detect (by tracking customer behavior) that someone is at risk of churn and then generate a customized offer that it sends to that person’s preferred communication channel to retain their loyalty – all without human intervention. And Agentforce Assistant would take these actions while maintaining privacy and security through Salesforce’s product-einstein-trust-layer.

Einstein 1 Editions provide organizations in every industry access to the best of Salesforce technology — including CRM, Agentforce Assistant, Data Cloud, Slack, and Tableau — in a single offering, helping them transform their business and deliver stronger customer experiences. Customers can additionally access Agentforce Assistant through Einstein add-on offerings to Enterprise and Unlimited Editions.

Customers looking to scale Agentforce Assistant can also invest in the new AI Implementation Bundle, which includes discounted sandbox environments and reduced-cost licenses for customizing and deploying conversational AI.

Agentforce Assistantt Sales actions, Service actions, and Copilot Analytics are generally available today.

• Read the news announcing Agentforce Assistant’s arrival and watch the video
• See how Salesforce is applying more than a decade of AI research to usher in a new era of productivity
• Read why Salesforce’s leading AI minds say trust is job number one for developing AI