Deploying Agentforce in Slack: Salesforce’s Journey to Saving over 500K Employee Hours Annually

When Salesforce integrated AI agents into Slack in February 2025, the results weren't immediate. But, following six months of testing and honing the agents, 86% of Salesforce employees lean on Agentforce to get answers and take action in Slack.

Slack isn't just where employees spend their day. It's where the conversations, decisions, and knowledge of an organization live. By bringing AI-powered agents into Slack, Salesforce ensures employees can take action in the flow of work and draw on that rich conversational context. This makes agents not only easier to access, but also smarter, more accurate, and more impactful.

Agentforce in Slack lets employees @ mention an autonomous agent with questions or actions to take. Early on, adoption varied greatly across departments. As employees began to explore the new capabilities, usage steadily grew. The feedback from those first experiences offered a clear view of what people truly needed-and those insights shaped how Salesforce refined the experience going forward.

Now, Agentforce is on its way to saving 500,000 hours annually across the business.

Why it matters: As "Customer Zero" for Agentforce in Slack, Salesforce's experience showed how enterprise AI agent adoption unfolds - complete with ‌ trial-and-error. Rolling out the capability internally led to critical insights about user behavior, agent design, and change management - which helped improve the product and create a tested playbook to guide other companies through their agentic transformation.

Lessons learned: Salesforce found that success depends as much on change management as on technical implementation. Key lessons included:

• Start simple, then expand: Specialized agents have been key to defining agent use cases and performance. Launching first with HR and IT use cases gave teams space to learn and test the tools before expanding into more specialized functions.
• User feedback shaped the roadmap: Employees wanted clarity on what agents could and couldn't do - and easy ways to escalate to a human when needed. That input influenced how agents were designed and introduced.
• Agents are only as good as their source data: At launch, agents pulled from a 2,000-article knowledge base - but quantity didn't equal quality. For example, BaseCamp agent risked giving conflicting guidance if outdated policies or redundant articles were included. Tightening the knowledge base to 600 high-quality, vetted resources improved accuracy and reliability.
• Pace of adoption may vary: Some departments jumped in right away. Others took longer to get comfortable with using agents. That learning curve shaped the rollout strategy. Adoption grew fastest when the first agent tackled tasks everyone encountered, like password resets, delivering quick wins. Once familiarity set in, it was easier to roll out team-specific agents.

What is Agentforce in Slack?: Agentforce isn't a single bot - it's a platform for building specialized AI agents that can be incorporated into Slack to help different teams get their work done faster and smarter. Together, these agents are becoming digital teammates, saving hundreds of thousands of hours while helping employees focus on higher-value, more meaningful work. Some of the agents used at Salesforce include:

• Sales Agent pulls pipeline data, forecasts, and deal context without leaving Slack, and can generate executive briefs, quiz sellers, or source competitive intel from across the business. It's projected to save an estimated 203,000 hours annually - time sales teams can redirect toward customers.
• BaseCamp Agent is a 24/7 workplace concierge. Some 25,000 Salesforce employees - over a third of the company - use it for everything from HR questions to procurement workflows, tailored by role and department.
• Techforce Agent automates IT tasks like password resets, device provisioning, and access requests - freeing IT to focus on bigger projects. It's already handled 64,000+ requests, returning 17,000 hours to technical teams for higher-priority work.
• Engineering Agent surfaces answers buried in docs, emails, and message threads, guiding employees through workflows instead of just pointing to content. It's projected to save 275,000 hours annually - the equivalent of 130 full-time engineers.

What they're saying:

• "Being Customer Zero means learning in real time. Our rollout of Agentforce in Slack revealed the real-world dynamics of AI adoption and the value of building solutions that work for teams, not just systems. The result: hundreds of thousands of hours reclaimed for higher-impact work." - Andy White, SVP of Technology, Salesforce
• "The Sales Agent in Slack has been a game changer for me. I used to spend a lot of time tracking down documents and answers on pricing, or competitive intel, or past customer meetings. Now I just ask the agent and get what I need instantly. It's saved me hours each week and lets me stay focused on selling and building relationships with customers." - Tyler Wilks, Account Executive, Salesforce
• "Collaboration is super important for us, and Slack has always been at the center of how we collaborate. reMarkable is all about focused work time, and being able to create workflows directly from Slack, with the help of agents who are creating those perfect workflows, that's something I'm super excited about." - Nico Cormier, CTO, reMarkable

Our rollout of Agentforce in Slack revealed the real-world dynamics of AI adoption and the value of building solutions that work for teams, not just systems.

Learn more: