Salesforce Agentforce 3: Inside the New Era of Autonomous AI Agents for Business

Salesforce Agentforce: What You Need to Know

Salesforce Agentforce marks a shift from generative AI to agentic AI, emphasizing intelligent agents that act independently to improve customer experience and service. Since its debut at Dreamforce 2024, Agentforce has reshaped how businesses interact with AI in customer support, marketing, and sales. Salesforce’s CEO Marc Benioff called it “the third wave of AI,” highlighting its low hallucination rate and ability to actively drive customer success.

What Is Agentforce?

Agentforce is a platform within the Salesforce ecosystem featuring AI-powered bots that don’t just respond—they take action based on internal data and large language models (LLMs). Unlike typical generative AI like ChatGPT, these agents can operate autonomously within a company’s parameters. Agentforce uses three main tools—Agent Builder, Model Builder, and Prompt Builder—to create customizable bots for various industries using low-code tools.

Initial Agentforce bots include:

• Campaign Optimizer: Manages full campaign lifecycles.
• Service Agent: Replaces traditional chatbots for customer service.
• Buyer: Assists B2B customers with product discovery, purchases, and tracking.
• Personal Shopper: Offers product recommendations and search help.
• Merchant: Supports merchandisers with promotions, product descriptions, and insights.
• Sales Development Representative: Engages leads 24/7.
• Sales Coach: Trains sales teams and helps sellers practice pitches.

Agentforce One became generally available in October 2024, with pricing starting at $2 per conversation and discounts for high-volume users. The Agentforce Partner Network enables third parties to deploy custom agents and actions.

How Agentforce Works

Agentforce is built for users without deep AI expertise. It’s powered by the Atlas Reasoning Engine, an upgraded version of Salesforce’s Einstein AI that simulates human planning and thinking. It adapts to real-time data and can operate autonomously within company guardrails. This makes AI accessible “for the rest of us,” allowing businesses to deploy intelligent agents without needing specialized knowledge in LLMs.

Agentforce Integrations with Salesforce Tools

Data Cloud

Data Cloud connects and harmonizes customer data, making it easier for Agentforce bots to deliver precise, context-aware responses. It supports structured and unstructured data, including emails and documents, and features retrieval-augmented generation (RAG) for uploading files and URLs as data sources.

Customer 360

Agentforce leverages Customer 360 apps across sales, service, marketing, and commerce for a unified customer view. This ensures seamless hand-offs and rich conversation histories.

Heroku

Heroku Managed Inference and AppLink simplify deploying and maintaining custom MCP servers, allowing developers to bring trusted, custom actions into Agentforce using secure infrastructure and DevOps automation.

MuleSoft

MuleSoft extends Agentforce by enabling API integrations and data sharing across platforms. New MCP connectors convert any API into an agent-ready asset with built-in security and traffic controls.

Slack

Slack lets employees interact with Agentforce directly—accessing information, triggering workflows, and managing tasks without switching apps. Agentforce also uses Slack Enterprise Search to learn from conversations and provide contextually relevant responses.

Tableau

Tableau Einstein integrates AI-powered visual analytics with Agentforce to embed data insights into workflows, helping employees make data-driven decisions.

Agentforce Testing Center

Launched in December 2024, the Testing Center enables businesses to verify that AI agents follow instructions, maintain accuracy, and perform efficiently before deployment. It offers tools for testing, auditing, and scaling agents to reduce operational risk.

Library of Skills and Integrations

This library offers ready-to-use skills and integrations across CRM, Slack, Tableau, and AppExchange partners. It simplifies building AI agents for common business functions, such as lead nurturing, sales coaching, and service automation.

Examples include:

• CRM Skills: Automate customer experiences, lead nurturing, and service scheduling.
• Sales Development: Create agents that nurture leads based on rules.
• Sales Coaching: Provide real-time feedback during sales calls.
• MuleSoft Integrations: Connect to ERP and marketing platforms for automating order fulfillment and personalized campaigns.

Agentforce for Retail

Announced in January 2025, Agentforce for Retail supports retail brands with pre-built agent skills for “digital labor.” These include order management, guided shopping, appointment scheduling, and loyalty promotion creation. Salesforce also introduced Retail Cloud with Modern POS to unify online and offline inventory data.

Companies Using Agentforce

Salesforce uses Agentforce internally, with agents handling 83% of support queries independently and reducing human escalations by half shortly after implementation. Other notable users include RBC Wealth Management, Accenture, AAA Washington, PepsiCo, Wayfair, and more.

Salesforce and Google Partnership Expansion

Salesforce customers will soon build Agentforce agents using Google Gemini and deploy Salesforce on Google Cloud. This integration connects Agentforce agents to real-time data, news, and events via Google Search and BigQuery, enhancing contextual awareness and response accuracy.

For example, supply chain agents can track shipments, monitor inventory, and anticipate disruptions using live data like weather or port congestion. By the end of 2025, Google Gemini models will be available within Agentforce, adding multi-modal capabilities and faster reasoning.

Agentforce for Field Service

Launched in April 2025, Agentforce for Field Service automates scheduling and routine tasks for dispatchers and technicians. It integrates with existing systems to fill schedule gaps, troubleshoot issues, and summarize job reports. Voice commands and audio playback help technicians work hands-free, improving productivity and safety.

Agentforce for HR Service

In May 2025, Salesforce introduced Agentforce for HR Service, embedding AI agents into HR platforms. Employees can handle common HR tasks conversationally—like updating personal info, managing time off, tracking cases, and submitting expenses—without switching systems. The agents connect with HRIS and HCM systems and follow company policies.

Trust and Governance for Agentforce

Also announced in May 2025, Salesforce added governance, security, and compliance features to ensure trusted AI deployment. These include:

• Salesforce Backup & Recover: Automated backups and restoration.
• Salesforce Archive: Compliance with data retention policies.
• Data Mask and Seed: Secure sandbox data handling.
• Shield 2.0: Enhanced user activity monitoring and threat detection.
• Security Center 2.0: Streamlined security posture management.

These tools provide visibility, control, and transparency throughout agent deployment.

New Pricing Models for Agentforce

Salesforce introduced flexible pricing options in May 2025, including:

• Flex Credits: Pay per action performed by agents, starting at 20 credits per action (equivalent to $0.10).
• Flex Agreement: Shift investments between human and digital labor licenses to match business needs.
• User Licenses and Add-ons: Unlimited employee-facing agent use with per-user, per-month pricing.

Agentforce for Financial Services

Also in May 2025, Salesforce launched Agentforce for Financial Services, featuring role-based AI agents for tasks like investment review prep, replacing lost credit cards, and loan discovery. Agents operate within compliance rules and integrate seamlessly with Financial Services Cloud.

Templates include:

• Financial Advisor and Banker agents for meeting prep and follow-up.
• Banking and Insurance service agents handling routine requests.
• Digital Loan Officer guiding borrowers through loan options.

Marketing Cloud Next with Agentforce

In June 2025, Salesforce introduced Marketing Cloud Next, embedding autonomous AI agents to manage marketing campaigns end-to-end. Marketers can set goals, and AI agents independently execute campaigns, personalize interactions, and optimize results across channels like email and SMS.

Agentforce 3 and MCP Support

Salesforce’s Agentforce 3 release focuses on observability and control through the new Agentforce Command Center. This tool provides a centralized dashboard to monitor agent health, analyze interactions, and optimize performance with AI-driven recommendations.

Agentforce 3 also introduces native support for the Model Context Protocol (MCP), enabling agents to connect to any MCP-compliant server without custom coding. This opens access to enterprise tools and resources from partners like AWS, Google Cloud, IBM, and more, all governed securely.

Additional improvements include reduced latency and expanded geographic support, making Agentforce more responsive and widely available.

For those interested in expanding their AI skills relevant to platforms like Agentforce, exploring comprehensive AI training can be a practical next step.