What is Agentforce?
Agentforce is Salesforce's AI agent platform. Unlike traditional chatbots that follow rigid scripts, Agentforce agents use the Atlas Reasoning Engine to understand context, access your Salesforce data, make decisions and take actions autonomously — all within the boundaries you define.
Salesforce Agentic Maturity Model
Salesforce recommends an incremental approach to AI agent adoption. Each level generates value on its own and prepares you for the next. You don't need to run from day one — what matters is getting started.
Fixed Rules and Repetitive Tasks
Automation of repetitive tasks using predefined rules, with no reasoning or learning capabilities. Chatbots and co-pilots that retrieve information but don't autonomously recommend or execute actions.
A Flow that assigns cases to queues based on product type.
Information Retrieval Agents
Agents that assist humans by retrieving information and recommending actions. They pull data and suggest next steps, but require human approval for execution.
A Service Agent that queries the Knowledge Base and suggests the solution to the human agent, who decides whether to apply it.
Simple Orchestration, Single Domain
Agents that autonomously orchestrate low-complexity tasks in a siloed data environment. They perform actions independently within a single operational area.
A Lead Nurturing Agent that automatically qualifies leads and schedules meetings without human intervention, operating only within Sales Cloud.
Complex Orchestration, Multiple Domains
Agents that autonomously orchestrate multiple workflows with harmonized data across multiple domains. They handle intricate cross-departmental processes simultaneously.
Service + FSL + Sales agents with integrations to ERP and inventory, identity resolution for a 360 view, and centralized governance.
Multi-Agent Orchestration
Any-to-any agent operability across disparate stacks with agent supervision. Multiple agents from different systems collaborate without human intervention across organizational boundaries.
A service agent detects a recurring issue, triggers the FSL agent to schedule a visit, and the sales agent proposes an upgrade — all coordinated automatically.
Every organization has its own pace. What matters is starting with a concrete use case, validating results and evolving step by step.
Standard agents available by cloud
Each Salesforce cloud has its own agent model. These are the ones we implement in our Agentforce packages.
Agentforce Sales
Customer-facing2 agents
Lead Nurturing Agent
Nurtures leads 24/7 via email: answers product questions, handles objections and books meetings with sales reps autonomously.
Engagement Agent
Converts inbound leads with personalized conversations. Greets web visitors, answers questions, presents products and books meetings.
Employee-facing5 agents
Sales Coach
AI role-play to train sales reps. Simulates buyers and provides personalized feedback per opportunity across 5 sales cycle stages.
Prospecting Agent
Searches, prioritizes and maintains updated prospect lists based on ideal customer profile, intent signals and CRM + web data.
Pipeline Management Agent
Keeps the pipeline automatically updated: updates stage, next steps and close date based on recent sales activity.
Account Research & Meeting Prep
Researches accounts and generates pre-meeting briefs combining CRM data, external sources and previous conversations.
Quoting Agent
Generates accurate and compliant quotes using natural language, following catalog rules, pricing and Revenue Cloud permissions.
Partner-facing1 agent
Partner Success Agent
24/7 partner support: access to product specs, personalized enablement and co-selling assistance.
Agentforce Service
Customer-facing1 agent
Service Agent
Autonomous agent that resolves cases 24/7 in 25+ languages, omnichannel (web, chat, email, messaging). Configured by activating topics from the Asset Library.
Employee-facing3 agents
Service Rep Assistant
AI assistant for human agents in the console. Generates action plans, verifies warranties, recommends next best actions and creates Knowledge articles from resolved cases.
IT Service Agent
Autonomously resolves employee IT requests: password resets, system access, diagnostics. Operates in Slack and Microsoft Teams.
HR Service Agent
Conversational HR resolutions 24/7. Answers questions about policies, benefits and employee processes via Slack and internal portals.
Agentforce Field Service
Customer-facing1 agent
Scheduling Agent
Autonomous agent with its own template in Agent Builder. Customers book, reschedule and cancel appointments via web or messaging in natural language, 24/7 without human intervention.
Employee-facing1 agent + 4 AI features
Dispatcher Assistant
Conversational interface pre-integrated in the Dispatcher Console. Identifies schedule gaps, recommends appointments to fill them and reassigns technicians considering skills, parts and SLAs.
AI Features — Mobile App (Technician)
Pre-Work Brief
Pre-visit summary with customer context, history, assets and required parts. Audio playback and Siri. Customizable with Prompt Builder.
On-Site Troubleshooting
Multimodal guided diagnosis with photos and error codes. On-device LLM that works offline. Queries manuals, previous repairs and sensor data.
Post-Work Summary
Automatic work order closure with AI-generated summary. Technician can refine with natural language or Siri.
Voice to Form
Hands-free Data Capture form filling by voice. Technician speaks and fields are automatically completed with contextual accuracy.
The topics shown for each agent come from official Salesforce documentation. Salesforce continuously updates the catalog of agents, topics and actions in the Asset Library.
Implementation packages
Three packages designed to accompany you at each level of your Agentforce adoption.
Activate
1 standard agent, 1 topic, 1 channel — 100% out-of-the-box
Deploy your first AI agent with standard Salesforce configuration. Validate that Agentforce works in your operation with a concrete use case and results in weeks.
- 1 standard agent configured
- 1 standard topic (Asset Library)
- Standard actions included with the topic
- 1 channel (Web Chat or Slack)
- Guardrails and escalation configured
Accelerate
RecommendedUp to 2 agents, 3 standard topics, custom actions and Data Cloud
Scale from a pilot to agents that reason with context. Standard topics with custom actions for your business, unified data with Data Cloud and point-to-point integrations.
- Up to 2 agents configured
- Up to 3 standard topics
- Up to 5 custom actions
- Basic Data Cloud (up to 2 data streams)
- Up to 2 point-to-point integrations
Transform
ConsultativeUp to 3 agents, custom topics, integrations and enterprise governance
Autonomous agents that operate across multiple domains with external integrations, identity resolution and advanced governance. For companies that want to transform their operation with AI.
- Up to 3 agents configured
- Up to 5 standard topics + up to 1 custom
- Up to 10 custom actions
- Data Cloud (up to 4 data streams) + identity resolution
- Up to 4 point-to-point integrations
Detailed comparison
Select a package in the table to highlight its column and compare in detail.
Agents and topics
Actions and integrations
Data Cloud and channels
Testing and governance
Deliverables and support
General scope
How we implement Agentforce
Our proven methodology for taking AI agents to production.
Discovery & use case
We identify the highest-impact use case, define topics, actions and channels. We evaluate available data and licensing prerequisites.
Configuration & build
We configure agents, instructions, guardrails and escalation. We connect Data Libraries, Data Cloud and selected channels.
Testing & validation
We validate with Testing Center (topic selection, action execution, multi-turn). Tests with real users and instruction tuning.
Go-Live & hypercare
Progressive deployment, agent metrics monitoring, continuous topic and instruction tuning. Training and documentation.
Data Cloud: the context that powers your agents
Without Data Cloud, your agent only sees the data inside Salesforce. With Data Cloud, your agent accesses a unified view of your customer — data from CRM, ERP, website, apps and any external source — to make decisions with full context.
Why does it matter for Agentforce?
An agent with partial data gives generic answers. An agent with unified data understands who your customer is, what they've purchased, what issues they've had and what they need — before they ask.
Data Streams
Connect external sources (ERP, databases, APIs) so your agents access real-time data without replicating information.
Identity Resolution
Unify customer profiles from multiple sources. Your agent knows that John Smith from the CRM, jsmith@company.com from the ERP and ticket #4521 are the same person.
Data Libraries (RAG)
Agent grounding in your Knowledge Base, documents and external sources. Answers are based on real data, not generic model knowledge.
Unified Knowledge
Connect Confluence, SharePoint, Google Drive and other sources so the agent accesses internal documentation without migrating it.
Data Cloud in our packages
Accelerate includes basic setup (up to 2 read data streams) so your agents have external context. Transform expands to up to 4 data streams and includes identity resolution for a real 360 customer view.