What are the Key Features of Salesforce Agentforce? Posted by
MacLain JohnsonThe short answer: Agentforce is Salesforce’s platform for building AI agents: assistants that answer questions and complete tasks (updating records, booking meetings, resolving cases) using your CRM data. At Dreamforce 2025 it became the umbrella brand for the whole Salesforce platform as Agentforce 360, which added Agentforce Builder, Agent Script, and voice support.
Agentforce has moved faster than any Salesforce product in recent memory: new name, new builder, new capabilities, all within about a year. This guide covers what Agentforce actually is, its key features as of mid-2026, and what to check before rolling it out.
What is Salesforce Agentforce?
Agentforce is the platform for creating, deploying, and managing AI agents inside Salesforce. Unlike a chatbot that follows a fixed script, an Agentforce agent interprets a request, retrieves relevant data from your org, decides which of its permitted actions applies, and executes it: drafting a reply, updating an opportunity, scheduling service, or escalating to a person. Agents work for customers (on your website, over messaging, or by phone) and for employees (inside Salesforce and Slack).
A naming note: at Dreamforce 2025 Salesforce reorganized its portfolio under the Agentforce 360 brand, which now spans the Agentforce 360 Platform, Data 360 (formerly Data Cloud), the Customer 360 apps, and Slack. Our guide to the Salesforce product name changes maps all the renames.
What are the key features of Agentforce?
- Agentforce Builder. The workspace for creating and testing agents, introduced with Agentforce 360. Admins can build visually on a no-code canvas, and developers can work in Agent Script, a scripting language that makes agent behavior precise and repeatable. Simulation and step-by-step trace debugging show exactly why an agent did what it did.
- Topics and actions. Agents are organized around topics (categories of work, like “order support”) containing actions (the specific things an agent may do). Actions can invoke Flows, Apex, prompt templates, or MuleSoft APIs, and Salesforce ships more than 200 prebuilt templates, topics, and actions across sales, service, marketing, and commerce.
- Grounding in your data. Agents draw on CRM records, knowledge articles, and Data 360 (formerly Data Cloud) profiles, so answers reflect your business rather than generic model knowledge. The Intelligent Context capability extends this to messy unstructured content.
- Agentforce Voice. Agents can now take phone calls with natural, low-latency speech while executing the same actions, with support for contact center platforms including Amazon Connect, Five9, Genesys, NiCE, and Vonage.
- Guardrails and observability. Instructions and permission boundaries limit what an agent may do, the Einstein Trust Layer governs how data reaches language models, and Agentforce Observability logs every session with quality scores and analytics so you can find and fix weak spots.
- Slack as the front door. For employees, agents live where the conversation already happens: Slack channels and DMs, plus the Salesforce UI itself.
What should you have in place before deploying Agentforce?
Agents are only as good as what they are grounded in. The prerequisites we check before any Agentforce build: clean, deduplicated CRM data; a knowledge base that actually answers your common questions; clear escalation paths to humans; and agreement on which actions the agent may take unattended versus draft-for-review. Most disappointing agent pilots trace back to skipping one of these, not to the technology.
CloudMasonry builds Agentforce assistants grounded in client data as part of our Salesforce consulting practice. If you are weighing where an agent would genuinely pay off, talk to our team, or read our take on the best use cases for Agentforce.
Frequently Asked Questions
What is the difference between Agentforce and Einstein?
Einstein is Salesforce’s long-standing brand for AI features like scoring and predictions, while Agentforce is the platform for AI agents that hold conversations and complete multi-step tasks. Under the Agentforce 360 brand, the earlier Einstein Copilot assistant evolved into Agentforce, and Einstein AI capabilities now power Agentforce behind the scenes.
Is Agentforce included with Salesforce licenses?
No. Agentforce is licensed separately from core Salesforce user licenses, and the licensing options change frequently. Confirm what is currently available with Salesforce or your implementation partner.
Do I need Data Cloud (Data 360) to use Agentforce?
Not for simple use cases: agents can work from CRM data and knowledge articles alone. For agents that need a complete customer picture across systems, Data 360 is the grounding layer Salesforce designed for the job, and most larger Agentforce deployments implement both together.
What is Agentforce Builder?
Agentforce Builder is the workspace introduced with Agentforce 360 for creating, testing, and debugging agents. It offers a no-code canvas for admins and Agent Script for developers who want precise, code-level control of agent behavior, plus simulation and trace debugging before anything reaches production.
MacLain JohnsonThe short answer: Agentforce is Salesforce’s platform for building AI agents: assistants that answer questions and complete tasks (updating records, booking meetings, resolving cases) using your CRM data. At Dreamforce 2025 it became the umbrella brand for the whole Salesforce platform as Agentforce 360, which added Agentforce Builder, Agent Script, and voice support.
Agentforce has moved faster than any Salesforce product in recent memory: new name, new builder, new capabilities, all within about a year. This guide covers what Agentforce actually is, its key features as of mid-2026, and what to check before rolling it out.
What is Salesforce Agentforce?
Agentforce is the platform for creating, deploying, and managing AI agents inside Salesforce. Unlike a chatbot that follows a fixed script, an Agentforce agent interprets a request, retrieves relevant data from your org, decides which of its permitted actions applies, and executes it: drafting a reply, updating an opportunity, scheduling service, or escalating to a person. Agents work for customers (on your website, over messaging, or by phone) and for employees (inside Salesforce and Slack).
A naming note: at Dreamforce 2025 Salesforce reorganized its portfolio under the Agentforce 360 brand, which now spans the Agentforce 360 Platform, Data 360 (formerly Data Cloud), the Customer 360 apps, and Slack. Our guide to the Salesforce product name changes maps all the renames.
What are the key features of Agentforce?
- Agentforce Builder. The workspace for creating and testing agents, introduced with Agentforce 360. Admins can build visually on a no-code canvas, and developers can work in Agent Script, a scripting language that makes agent behavior precise and repeatable. Simulation and step-by-step trace debugging show exactly why an agent did what it did.
- Topics and actions. Agents are organized around topics (categories of work, like “order support”) containing actions (the specific things an agent may do). Actions can invoke Flows, Apex, prompt templates, or MuleSoft APIs, and Salesforce ships more than 200 prebuilt templates, topics, and actions across sales, service, marketing, and commerce.
- Grounding in your data. Agents draw on CRM records, knowledge articles, and Data 360 (formerly Data Cloud) profiles, so answers reflect your business rather than generic model knowledge. The Intelligent Context capability extends this to messy unstructured content.
- Agentforce Voice. Agents can now take phone calls with natural, low-latency speech while executing the same actions, with support for contact center platforms including Amazon Connect, Five9, Genesys, NiCE, and Vonage.
- Guardrails and observability. Instructions and permission boundaries limit what an agent may do, the Einstein Trust Layer governs how data reaches language models, and Agentforce Observability logs every session with quality scores and analytics so you can find and fix weak spots.
- Slack as the front door. For employees, agents live where the conversation already happens: Slack channels and DMs, plus the Salesforce UI itself.
What should you have in place before deploying Agentforce?
Agents are only as good as what they are grounded in. The prerequisites we check before any Agentforce build: clean, deduplicated CRM data; a knowledge base that actually answers your common questions; clear escalation paths to humans; and agreement on which actions the agent may take unattended versus draft-for-review. Most disappointing agent pilots trace back to skipping one of these, not to the technology.
CloudMasonry builds Agentforce assistants grounded in client data as part of our Salesforce consulting practice. If you are weighing where an agent would genuinely pay off, talk to our team, or read our take on the best use cases for Agentforce.
Frequently Asked Questions
What is the difference between Agentforce and Einstein?
Einstein is Salesforce’s long-standing brand for AI features like scoring and predictions, while Agentforce is the platform for AI agents that hold conversations and complete multi-step tasks. Under the Agentforce 360 brand, the earlier Einstein Copilot assistant evolved into Agentforce, and Einstein AI capabilities now power Agentforce behind the scenes.
Is Agentforce included with Salesforce licenses?
No. Agentforce is licensed separately from core Salesforce user licenses, and the licensing options change frequently. Confirm what is currently available with Salesforce or your implementation partner.
Do I need Data Cloud (Data 360) to use Agentforce?
Not for simple use cases: agents can work from CRM data and knowledge articles alone. For agents that need a complete customer picture across systems, Data 360 is the grounding layer Salesforce designed for the job, and most larger Agentforce deployments implement both together.
What is Agentforce Builder?
Agentforce Builder is the workspace introduced with Agentforce 360 for creating, testing, and debugging agents. It offers a no-code canvas for admins and Agent Script for developers who want precise, code-level control of agent behavior, plus simulation and trace debugging before anything reaches production.
Client Success Stories
Beamery
A fast-growing candidate relationship management software company. Read how CloudMasonry delivered Sales Cloud, Service Cloud, Tableau, and Einstein Analytics.
Kensington
The worldwide leader in laptop security locks, a division of ACCO Brands. See how CloudMasonry delivered Sales Cloud, Pardot, and Einstein Analytics.
Gradient AI
An AI company helping insurers predict underwriting and claim risk with greater accuracy. Read how an AI-first business runs its own operations on Salesforce.
Packback
A Chicago education technology company built around an AI platform. See how CloudMasonry integrated Salesforce Sales Cloud with their core product.