What is Salesforce Agentforce?
Salesforce Agentforce is an enterprise AI agent platform for organisations that want AI agents to take action across CRM, service, sales, marketing, commerce, and internal workflows. It is designed for teams already using Salesforce that need agents connected to customer records, business rules, knowledge, and approved actions rather than a standalone chatbot.
Agentforce supports the lifecycle of building, testing, deploying, managing, and observing agents. Salesforce positions it as a platform for customer-facing and employee-facing agents that can use Salesforce data and connected external systems with governance controls.
Agentforce platform capabilities
- Low-code and pro-code building: Agentforce Builder, visual workflows, prompt tools, Flows, Apex, and MuleSoft APIs help teams configure agents and actions.
- CRM and enterprise context: agents can use Salesforce records, knowledge, metadata, and connected structured or unstructured data. Salesforce describes retrieval-augmented generation and vector capabilities for grounding responses in relevant business context.
- Workflow orchestration: agents can combine deterministic workflow steps with language-model reasoning, then take governed actions across connected systems.
- Channels and use cases: deployments can support customer service, sales development, lead nurturing, employee workflows, chat, voice, apps, and portals.
- Governance and observability: Salesforce provides controls for testing, monitoring, security, access, and operational oversight.
Salesforce Agentforce pricing
Salesforce lists Agentforce Foundations at US$0. Its consumption options include Flex Credits at US$500 per 100,000 credits, with a standard action consuming 20 credits and a voice action consuming 30 credits, plus conversation pricing at US$2 per conversation. Agentforce add-ons are US$125 per user per month, Industries add-ons are US$150, and the Agentforce User Licence is US$5 per user per month plus Flex Credits.
Salesforce's new bundled Agentforce Sales and Service editions are Core at US$195 per user per month, Advanced at US$395, and Max at US$550. Salesforce says these editions combine AI with Slack, analytics, security, and support; Max keeps the former Agentforce 1 price. New Industry Edition pricing is due later in the northern-hemisphere fall, while existing legacy-edition customer pricing is unchanged.
Models, data, and integrations
Agentforce is an orchestration platform rather than a single-model product. Its value is in combining language-model reasoning with Salesforce business context, retrieval, workflow logic, and action execution. Salesforce also documents support for open integration approaches including Model Context Protocol (MCP) and agent-to-agent (A2A) patterns, alongside Salesforce-native tools such as Flow, Apex, MuleSoft, and Data 360.
For an evaluation, ask which models and features are available in your Salesforce edition and region, what data is available to each agent, how actions are approved and logged, and how usage is measured before rollout.
Best fit for Salesforce Agentforce
Agentforce is strongest for organisations with meaningful Salesforce adoption that want to automate sales, service, commerce, or employee processes with access to CRM data and established business workflows. It is less compelling for a small team that only needs a generic website chatbot or does not rely on Salesforce as a core system of record.
Evaluation checklist
- Start with a narrow, measurable workflow such as case triage, lead qualification, knowledge answers, appointment scheduling, or sales coaching.
- Map every agent action, system permission, escalation path, and human handoff before production deployment.
- Estimate Flex Credit usage from expected actions and volumes; monitor costs with the relevant Salesforce usage tools.
- Test response grounding, data access, guardrails, and failure cases with realistic customer and employee scenarios.