Salesforce’s AI agents are smarter, more transparent, and ready to transform workflows. Here’s what businesses need to know—and what’s next.
Just months after introducing Agentforce, Salesforce is back with Agentforce 2.0. At the launch event, Marc Benioff and Salesforce’s product leaders doubled down on a bold vision: AI agents are the future of work, and Agentforce is the Digital Labor Platform.
2.0 so quickly raises a natural question: Why so soon? The answer is simple: AI isn’t slowing down, and neither is Salesforce. Agentforce 2.0 isn’t a rebrand; it’s a demonstration of Salesforce’s continued investment in making AI more useful for businesses.
For businesses, this update offers real opportunities—but also real questions. What can these agents actually do today? What’s needed to ensure Return on AI (ROAI)? And where does Salesforce’s vision for digital labor leave your team today?
At Coastal, we’re cutting through the noise. This blog dives into:
- Why Agentforce 2.0 matters.
- What’s new and what businesses need to know before adopting it.
- How to think strategically about AI agents—and where they’re headed next.
If you’re wondering how to make AI agents work for you, this is your guide to what Agentforce 2.0 means—and how to stay ready for what’s next.
What’s New in Agentforce 2.0
The updates to Agentforce focus on three critical areas: smarter decision-making, simpler agent creation, and deeper integration into workflows.
For many businesses, these changes remove barriers to adoption—making AI agents easier to test and launch. But for more advanced teams already experimenting with AI, Agentforce 2.0 might feel like it stops short of the bleeding edge. And that’s not necessarily a problem.
1. Smarter Decision-making: Agents with Atlas Reasoning
At the heart of Agentforce 2.0 is the Atlas Reasoning Engine, which makes agents smarter, more reliable, and—critically—more transparent.
How It Works
Agents don’t stop at surface-level answers—they ‘reason’ through tasks step by step. For instance: ‘How will falling interest rates impact a portfolio?’
Atlas doesn’t stop at a simple trend report. It can:
- Pull live interest rate trends.
- Analyze the customer’s portfolio details.
- Cross-check internal policies to ensure compliance.
- Deliver a clear, actionable answer.
What sets Atlas apart is its citations—showing teams where the data came from and how the answer was built. This directly addresses one of AI’s biggest challenges: hallucinations.
Why It Matters: Trust is the foundation for AI adoption. Atlas provides transparency, letting teams validate outputs, reduce errors, and build confidence that agents are doing their job well.
2. Simpler Agent Creation: Natural Language Agent Building
Agentforce 2.0 lowers the barrier to AI creation with plain language prompts. Instead of writing code or navigating complex workflows, you describe what you need, and Agentforce builds the draft.
How It Works
Say, “I need an agent to pre-qualify job candidates and schedule interviews.” Agentforce scans your Salesforce data, flows, and processes to generate suggestions or build skills—modular sets of instructions agents use to complete tasks.
- Sales Skills: Qualify leads, summarize pipelines, schedule follow-ups.
- Service Skills: Manage tier-one support, escalate cases, pull customer histories.
- Custom Skills: For unique processes, teams can define new skills tailored to their workflows.
These agents don’t go straight to production. Teams validate and refine them using tools like Testing Center, ensuring they perform reliably.
Why It Matters: Simplified creation makes AI agents more accessible—but “easy to build” doesn’t mean “ready to use.” Testing and governance are still critical for scaling success.
Coastal CEO Eric Berridge building an agent at Agentforce World Tour NYC.
3. Deeper Integration into Workflows: Slack as the AI Work Hub
Salesforce is turning Slack into the central hub for AI agents, embedding them directly into where teams collaborate and get work done.
What Agents Can Do in Slack
- Surface Critical Insights: Ask, “What did I miss on the Jones account?” Agents provide meeting summaries, relevant documents, and next steps.
- Automate the Mundane: Agents can schedule tasks, summarize conversations, or update CRM records without prompting.
- Connect Systems: With MuleSoft connectors, agents can interact with tools like SAP, Workday, or ERPs—processing refunds, pulling inventory, or managing HR workflows.
Why It Matters: Many AI tools struggle with adoption. By meeting teams where they already work, Agentforce removes friction and makes AI feel natural—like a teammate, not another tool.
Data Cloud: Enhancing Agentforce 2.0
A key part of Salesforce’s Agentforce strategy is using Data Cloud as a tool to enhance agent performance. By unifying structured and unstructured data into a single platform, Data Cloud is designed to give agents the full context they need to make smarter, more consistent decisions.
Salesforce’s emphasis on Data Cloud highlights their push to address one of AI’s biggest challenges: disconnected or incomplete data. With real-time integration, agents can draw on up-to-date information from CRM records, call logs, and external data sources, reducing errors and improving reliability. Salesforce presents this as a way to streamline workflows and deliver more value for businesses leveraging their platform.
However, while Data Cloud is a powerful tool for those with siloed or complex data systems, it’s not required to adopt Agentforce. Businesses with well-integrated data sources can still benefit from AI agents without needing the added layer of Data Cloud. For organizations already struggling with disconnected systems, though, Data Cloud may be an opportunity to amplify what Agentforce can deliver.
The Takeaway: Data Cloud strengthens Agentforce’s capabilities by solving key data challenges, but its value depends on the specific needs and infrastructure of each business. Salesforce positions the two as complementary, offering a roadmap for organizations looking to unify their data and scale AI adoption.
The Reality Check: What Businesses Need to Know
Salesforce’s success—like the reported 50% reduction in escalations on Agentforce-powered help.salesforce.com—is a promising step. But the truth is, AI agents still have a way to go.
For teams using Agentforce, the reality can feel less polished:
- Agents sometimes provide inconsistent outputs.
- Scaling agents across messy or disconnected data sources remains a challenge.
This doesn’t mean Agentforce 2.0 isn’t valuable. It means businesses need to deploy it thoughtfully and realistically.
1. Data Quality Defines Success
AI agents depend on clean, structured, and reliable data. Inconsistent knowledge bases or disjointed repositories (like help articles mixed with developer docs) can confuse agents and lead to inconsistent answers.
What Businesses Can Do: Start small. Focus agents on well-defined, clean datasets to build reliability before scaling up.
2. Trust Still Requires Work
Citations build trust, but they’re not a silver bullet. Agents can still deliver incomplete or outdated answers, particularly in complex setups.
What Businesses Can Do: Treat agents like a work in progress. Validate performance with a testing center and build iterative feedback loops to improve results over time.
3. Humans Are Still Essential
Agentforce doesn’t replace human expertise—it scales efficiency. Teams are still needed to oversee, validate, and guide agents, especially in nuanced or complex workflows.
What Businesses Can Do: Deploy agents where they add immediate value—like tier-one support, scheduling, or data retrieval—and keep humans in the loop for oversight and decision-making.
4. What to Know About Pricing
Salesforce has confirmed that Agentforce 2.0 retains its $2/conversation pricing model, continuing the shift toward a consumption-based approach rather than a traditional subscription. For businesses, this means costs will scale with usage—making careful planning around deployment and use cases even more critical.
Where Agentforce is Headed Next
Salesforce’s vision for Agentforce stretches beyond today’s capabilities. As AI evolves, so will the potential of digital labor:
- Voice Interfaces: As demonstrated at Dreamforce earlier this year, Salesforce is working on future capabilities where agents interact with users via voice—not just text—opening up new channels for hands-free support.
- Image and Video Interpretation: Agents may soon analyze visual data, creating opportunities for tasks like identifying equipment issues in field service or processing visual content for customer support.
- Broader Horizons: Framing Agentforce as a “digital labor platform” signals Salesforce’s push to reframe how businesses think about AI—less as software and more as a scalable workforce. Down the line, this vision could extend AI agents into entirely new territories, including physical robotics and automation tools that move beyond the digital workspace.
Salesforce has set an ambitious roadmap for Agentforce. The platform will reach General Availability (GA) in February 2025, giving businesses full access to its enhanced capabilities. Looking further ahead, Salesforce has teased Agentforce 3.0, slated for May 2025, which promises to expand on this foundation with even more advanced features.
These advancements won’t happen overnight. But businesses that focus on readiness today—testing use cases, improving data quality, and scaling thoughtfully—will be in the best position to take advantage of what comes next.
What Businesses Should Do Now
Meaningful adoption starts with a smart, practical approach:
- Start with High-Impact Use Cases: ROAI—Return on AI—isn’t just about the technology itself. It’s about integrating AI into your business strategy, identifying areas where it can make the most immediate impact, and embedding change management into the process. Focus on use cases where agents can deliver clear value today—like streamlining scheduling or supporting sales teams with lead qualification.
- Prepare Your Data: Clean, connected data is the foundation for reliable AI. If your systems are siloed or your knowledge sources are scattered, prioritize fixing that first.
- Test, Validate, and Iterate: Use tools like Salesforce’s Testing Center to test agent performance before going live. Build feedback loops to refine outputs and improve results over time.
Take the Next Step with Coastal’s Agentforce Workshop
With the right strategy and foundation, businesses can deploy AI agents that drive meaningful impact, reduce manual effort, and free teams to focus on the work that matters most.
At Coastal, we’re here to help you lay the groundwork:
- Pinpointing where AI can deliver the most value for your organization.
- Assessing your data readiness to ensure AI agents have the right inputs to perform effectively.
- Exploring use cases and setting a strategy to move from curiosity to actionable next steps.
Our complimentary Agentforce Workshop helps you find where AI can add value, get your data ready, and set a clear, practical plan for success.
Ready to make Agentforce work for you? Sign up for a workshop.
Fill out the form, and we’ll reach out to schedule a time. Complimentary workshops are available for qualified users, with pre-qualification required.
