Saleforce Agentforce

Salesforce Agentforce is an enterprise agentic AI platform designed to help organizations create autonomous AI agents that can perform real business tasks rather than simply answer questions. Agents can understand requests, retrieve relevant company information, determine appropriate actions, and execute multi-step workflows across sales, customer service, commerce, marketing, and internal operations. Agentforce is deeply integrated with the broader Salesforce platform, allowing agents to work directly with CRM data and existing business processes.

At the center of Agentforce is the Atlas Reasoning Engine, which breaks complex requests into smaller tasks and continuously evaluates what action should happen next. It combines agentic reasoning with advanced retrieval-augmented generation (RAG), structured and unstructured enterprise data, topic classification, and iterative reasoning loops. This enables an agent to move beyond generating responses and perform actions such as updating customer records, resolving service cases, scheduling appointments, or executing workflows.

Organizations create agents through Agentforce Builder, Salesforce's low-code development environment. Teams can start from prebuilt agents and templates or create custom agents using natural-language instructions. Agents can connect to Salesforce Flows, Apex, MuleSoft APIs, prompt templates, knowledge articles, external systems, and enterprise data. Developers can apply guardrails and deterministic business logic, while built-in testing and observability tools help organizations evaluate agent behavior before and after deployment.

Agentforce supports customer-facing and employee-facing experiences across chat, voice, mobile applications, portals, and other business channels. Agentforce Voice provides conversational voice interactions, while Agentforce Mobile SDKs enable organizations to embed agents into native iOS and Android applications. The platform also supports Model Context Protocol (MCP), allowing agents to connect with external systems and tools through an open integration standard.

Salesforce also provides Agentforce Observability for monitoring agent reasoning, task completion, performance, and operational impact. Agent Script gives developers deeper control by combining deterministic workflow logic with flexible LLM reasoning, while batch testing helps teams simulate scenarios and perform regression testing before deploying changes at scale.

Agentforce uses several pricing approaches. Salesforce Foundations provides core Agentforce capabilities at $0 for eligible customers. Consumption pricing includes $500 per 100,000 Flex Credits, while conversation-based pricing is $2 per conversation. Salesforce also lists Agentforce add-ons at $125/user/month, an Agentforce User License at $5/user/month plus Flex Credits, and Agentforce 1 Editions starting at $550/user/month.