Bringing agentic AI into your organization is a little more involved than flipping on a switch. This technology is changing the way employees and teams operate altogether — and laying the groundwork for AI agents is just as important as the agents themselves.
It’s about rethinking fundamental work processes: how decisions are made, actions are taken, and intelligence is applied in real time. The success of these agents depends less on how capable they are and more on how well the environment allows them to contribute meaningfully.
In traditional enterprise environments, software is siloed, workflows are fragmented, and context is hard to maintain. People know how to navigate this. They fill in the gaps with intuition. They escalate. They tap into tribal knowledge.
An AI agent, no matter how intelligent, cannot thrive in this ambiguity. They need environments that make context explicit, action immediate, and interaction fluid. If that world isn’t set up properly, you get the failed AI initiatives, wasted investment, and low adoption rates. In fact, a recent MIT report found that 95% of enterprise generative AI pilots fail to deliver ROI because of “brittle workflows, lack of contextual learning, and misalignment with day-to-day operations.”
This isn’t a new interface challenge. It’s a systemic one.
Principles for designing agentic work
Designing for agentic work means shifting the focus from individual interfaces to collaborative systems. It requires thinking not just about what an agent can do, but how that capability fits into the natural rhythms of team dynamics.
Key dimensions of this design shift include:
- Proximity to work: Agents should operate where decisions are being made, not in sidebars or dashboards.
- Access and authorization: Governed access to data must be fine-grained, real-time, and role-aware.
- Signals and handoffs: It should be clear when an agent takes initiative, when it waits for a person to approve, and when it defers.
- Recovery and escalation: Just like with human error, there must be lightweight paths to correct, clarify, or re-engage with an AI agent.
- Embedded feedback loops: Every interaction is an opportunity to learn, not just for the agent but for the system.
- Reduce cognitive load: Systems should be designed to minimize mental friction and support fluid interaction.
When an organization is designed around these principles, the result is more than a productive system. It’s a more intelligent one, where the value of AI is multiplied across people, actions, and decisions, and finally closes the gap between potential and performance.
What’s your agentic AI strategy?
Our playbook is your free guide to becoming an agentic enterprise. Learn about use cases, deployment, and AI skills, and download interactive worksheets for your team.
Slack and Salesforce: The operating system for the agentic enterprise
The journey to unlocking AI’s full potential doesn’t begin with better models. It begins with understanding that real enterprise work involves an ever-changing mosaic of people, conversations, decisions, and rhythms. The real power of AI is not in automating tasks, but in enriching the shared experience of work.
Traditional enterprise tools break up the context you need to drive work across dashboards, applications, and workflows. When AI is layered on as an afterthought, it never gets the full picture of your business that’s needed to be truly effective.
Slack fixes this by bringing together your team, applications, and data into a connected, open, AI-powered platform. Instead of operating in isolation, employee agents in Slack participate directly in your workflows. An agent can join a project channel, grasp the full context, and take action when appropriate. When it needs human input, the handoff is seamless because everything happens in the same collaborative space.
At Salesforce, we challenged ourselves to prove the model by becoming Customer Zero for using Slack as our agentic operating system for work. We’ve given employees digital teammates that can answer questions with your organization’s context, resolve issues, and automate repetitive work across functions. For example, a salesperson can ask Slackbot to summarize a recent meeting with a customer, identify next steps, and generate a polished follow-up email in mere seconds. We’re seeing many use cases improve our internal operations and outreach as well:
- Our Engineering Agent has handled 18,000 support interactions in just six months, projected to save 275,000 hours annually.
- Our Sales Agent, deployed in Slack where teams already work, is helping 25,000+ sellers save 203,000 hours a year with instant access to deal insights and competitive intel.
- In IT, this digital workforce has reduced our average case handle time by 35%, deflecting thousands of tickets per month with fast resolutions.
The results extend far beyond Salesforce. ReMarkable, a company creating a paper-like experience on a tablet, demonstrates how Agentforce transforms both customer and employee experiences. After reaching 3 million devices sold and a $1 billion valuation, reMarkable needed to scale smarter. They deployed “Mark,” an Agentforce-powered customer service agent that handles over 25,000 conversations with a 35% case deflection rate. “Saga,” their internal IT agent in Slack, instantly resolves password resets and Jira tickets without breaking employees’ creative flow.
“We’re building a digital workforce with Agentforce inside Slack. It’s freeing us up to collaborate, innovate, and move faster,” said Bettina Kotogany, system administrator at reMarkable. The company credits these AI teammates with fundamentally changing how their teams operate — fewer distractions, faster resolutions, and more time for meaningful work.
Similarly, Plative, a 200-person tech consulting firm, built three core Slack agents in under a month and saw immediate impact: 50% faster sales call prep time, 50% increase in upsell bookings, and the ability to avoid hiring one additional full-time employee for every five consultants.
“Our Agentforce Agents are like an octopus, with secure tentacles into Salesforce, Jira, Google Drive, and all our systems, and out to the best LLMs from OpenAI, Anthropic, Google, and Perplexity,” said Miftah Khan, SVP and head of professional services at Plative.
This is the promise of agentic AI: Agents harmonize with teams rather than interrupt them. Through this, we shift from fragmented interactions to fluid, co-created progress. From isolated brilliance to collective intelligence. Meeting this potential requires a connected, open, flexible platform capable of bringing together every facet of work.
To realize AI’s full potential, organizations need a connected, open, and flexible platform that can unify every aspect of work. Organizations that embrace this transformation will maintain their competitive edge, gaining the clarity needed to adapt and accelerate regardless of future AI developments.
The future of work is not defined by a new interface. It is through a more intelligent experience, built collectively, one conversation at a time.
Start using AI in Slack
Companies everywhere are using AI in Slack to work smarter, save time, and add some ease to their everyday lives. Learn more about meeting recaps, fast searching across the company, and using automated workflows.
