Salesforce Agentforce in Slack: AI Agents That Work Alongside Your Team

Salesforce Agentforce brings AI agents directly into Slack as team members. @Mention them in channels, assign tasks, and automate workflows across HR, IT, Sales, Service, and more. Here is everything you need to know.

June 22, 2026

Quick answer

Salesforce Agentforce is a platform that turns AI agents into Slack teammates. You @mention them in channels and DMs just like any coworker. They access conversational context from Slack and connected business systems to answer questions, automate tasks, and execute actions — all within Einstein Trust Layer security guardrails. Choose from pre-built templates for HR, IT, Sales, Service, Marketing, Legal, Engineering, or Supply Chain, or build custom AI assistants using the Slack API.

What Is Salesforce Agentforce?

Agentforce is Salesforce's platform for creating, deploying, and managing AI agents that operate alongside human employees. In Slack, these agents become active team members — they live in channels, participate in threads, respond to DMs, and take action autonomously within defined boundaries.

Unlike traditional chatbots that follow scripted decision trees, Agentforce agents use large language models, retrieval-augmented generation (RAG), and real-time conversational data to understand context, reason through problems, and execute multi-step workflows. They learn from the conversational history of your workspace and improve over time without manual retraining.

For teams already using Slack as their central hub, Agentforce eliminates the need to switch between tools. An agent can pull customer history from Salesforce, check inventory from an ERP, draft a response, and post it to a channel — all triggered by a single @mention.

The Three Types of AI Agents in Slack

Slack supports three categories of AI agents, each with different levels of customization and control:

1. Agentforce (Salesforce)

Pre-built agent templates powered by Salesforce. They draw on conversational data from Slack plus trusted business data from Salesforce and connected systems. Best for teams that want a ready-to-deploy solution with enterprise-grade security.

2. Custom AI Assistants (Slack API)

Build your own AI assistants using the Slack API and your preferred LLM provider. Full control over behavior, data sources, and integration logic. These assistants can be deployed as Slack apps with custom functions and triggers. Read our guide on developing AI agents with Bolt for Slack to get started.

3. Third-Party AI Assistants

Pre-configured AI apps from the Slack Marketplace by partners like Adobe Express and Cohere. These are ready to install and require no development work. They are reviewed by Slack for quality and security before publication.

Pre-Built Agent Templates for Every Department

Agentforce ships with ready-to-use agent templates tailored to common business functions. Each template comes with pre-configured skills, suggested data access scopes, and recommended Slack actions:

HR Agent

Assists with onboarding, benefits questions, and professional development guidance. Handles tier-1 HR requests so your people team can focus on strategic initiatives.

IT Agent

Resolves support tickets, runs incident communications, and handles access requests directly in Slack channels. Escalates only when human intervention is required.

Service Agent

Identifies subject matter experts, surfaces historical data, and resolves customer issues by pulling context from past conversations and connected CRM systems.

Sales Agent

Prepares executive briefs, creates proposals, and proactively reaches out to leads. Works alongside your sales team inside deal-specific Slack channels.

Marketing Agent

Optimizes campaigns, generates content drafts, builds campaign plans, and tracks performance metrics — all within Slack conversations.

Legal Agent

Simplifies compliance workflows, automates approval processes, and answers policy questions. Reduces the cycle time for legal reviews and contract approvals.

Engineering Agent

Accelerates planning, design, development, and QA cycles. Tracks bugs, surfaces PR status, and coordinates releases across engineering channels.

Supply Chain Agent

Manages supplier communications, automates warehouse operations, and improves demand forecasting by pulling data from ERP systems into Slack.

Why Conversational Context Makes Agents Smarter

The key differentiator of Agentforce in Slack is access to conversational data. Every Slack workspace is a living repository of institutional knowledge — decisions made in threads, context shared in channels, and expertise scattered across DMs. Agentforce agents tap into this unstructured data using retrieval-augmented generation (RAG) to produce relevant, context-aware responses.

When an agent is @mentioned with a question, it does not just search a database. It searches the conversational history of your workspace, pulls related messages and files, checks connected business systems via the MCP Slack server, and synthesizes everything into an answer or action. The result is an agent that understands context the way a human team member would — because it has access to the same conversations.

According to Slack, organizations that connect conversational data to their AI agents see up to a 40% improvement in task performance compared to agents that only access structured databases. Unstructured data — messages, files, audio, video — improves an agent's reasoning and decision-making capabilities significantly.

What Agentforce Agents Can Do in Slack

Once deployed, Agentforce agents can execute a growing set of Slack-native actions autonomously:

Create and manage channels

Agents can spin up new channels for projects, incidents, or initiatives — complete with the right members, topic, and purpose — without waiting for a human admin.

Update Slack canvases

Agents can create and update canvases with meeting notes, project status, or documentation. Your team always has access to the latest information inside Slack.

Send DMs and notifications

Agents can reach out to team members via direct message to request information, notify about changes, or escalate issues that require human intervention.

Manage lists and workflows

Agents can create and update Slack lists, trigger workflows, and track task progress. Combined with the Slack Workflow Builder, agents become powerful automation engines inside your workspace.

Security and Trust: The Einstein Trust Layer

Every Agentforce interaction is protected by the Einstein Trust Layer — a security framework built into the Salesforce platform that works silently behind every agent response:

No LLM training on your data

Salesforce never uses customer data to train large language models. After the LLM generates a response, all data is discarded through no-data-retention agreements with providers.

Data masking

Sensitive information — company names, person names, financial data — is masked before being sent to the LLM. The agent never processes raw sensitive data.

Toxicity monitoring

All agent outputs are continuously scored for toxicity and harmful content. Guardrails prevent agents from exhibiting unwanted behaviors or accessing restricted data.

For teams that need custom AI agents with specific security requirements, see our guide on connecting Hermes Agent to your Slack workspace for a detailed walkthrough of OAuth scopes, token storage, and channel restrictions.

FAQ

What is Salesforce Agentforce?

Agentforce is Salesforce's platform for building and deploying AI agents that work alongside humans. Inside Slack, Agentforce agents become team members that can be @mentioned, assigned tasks, and trusted to execute actions within defined guardrails.

How is Agentforce different from Slack's built-in AI?

Slack's built-in AI (like Slack AI search and recaps) focuses on helping humans find information faster. Agentforce agents are autonomous — they take action, make decisions within their scope, and execute multi-step workflows without waiting for human input at every step.

Can Agentforce agents access my private Slack channels?

No. Administrators control exactly which data and channels agents can access. By default, agents can only see public conversational data that you explicitly grant. Private channels and DMs are excluded unless specifically authorized.

Does Agentforce use my company data to train LLMs?

Absolutely not. Salesforce never uses customer data to train large language models. The Einstein Trust Layer ensures data masking, no-data-retention agreements with LLM providers, and continuous toxicity monitoring on all agent outputs.

Can I build custom AI agents alongside Agentforce?

Yes. Slack's platform supports three types of agents: Agentforce (pre-built templates), custom AI assistants built with the Slack API and your choice of LLM, and third-party assistants from Slack Marketplace like Adobe Express and Cohere.

Next steps

Salesforce Agentforce is transforming how teams collaborate with AI in Slack. Whether you use pre-built templates or build custom AI assistants, the key is starting with clear use cases and the right data access. To learn more about the underlying technology, read our comparison of Slack AI agent vs custom agents and our guide on what you get with a Slack AI agents subscription.

Need help deploying AI agents in Slack?

We build custom AI agents, configure Agentforce integrations, and automate Slack workflows for teams that want to move faster. Book a call and tell us what you need.