Salesforce Launches Agentforce 3 to Boost Visibility, Control and Scale

Salesforce today announced Agentforce 3, a major leap in its digital labour platform, designed to help organisations scale AI agents without compromising oversight or governance. The update includes a new Command Center for real-time observability, seamless interoperability with open standards like Model Context Protocol (MCP), and enhanced architecture for enterprise-grade trust, performance, and control.

Full Visibility and Observability with the Command Center

Agentforce 3 introduces a dedicated Command Center within Agentforce Studio — a unified observability layer allowing teams to monitor agent activity, track health metrics, and optimise outcomes in real time. With detailed analytics on latency, escalations, adoption, cost, and more, enterprises can fine-tune performance and scale AI usage with confidence. Integration with tools like Splunk, Datadog, and OpenTelemetry ensures compatibility with existing observability stacks.

“Agentforce autonomously resolved 70% of our administrative chats during peak tax season,” said Ryan Teeples, CTO at 1-800Accountant. “The new observability tools give us the clarity to optimise quickly and scale securely.”

Open Standards, Enterprise-Grade Security, and Trusted Ecosystem

Agentforce 3 includes built-in support for MCP, the emerging open standard likened to “USB-C for AI.” This allows plug-and-play connections to over 30 partners such as AWS, Google Cloud, Notion, Stripe, PayPal, and WRITER. MuleSoft now allows teams to turn any API into an MCP-compatible endpoint, while Heroku supports hosting and managing custom MCP servers securely.

“Our ability to adopt AI agents while staying compliant and in control has accelerated,” said Mollie Bodensteiner, SVP of Operations at Engine. “Salesforce’s open ecosystem and native MCP support are key enablers.”

Enhanced Architecture for Real-World Enterprise Needs

The revamped Atlas architecture delivers 50% lower latency, improved accuracy through web search and citations, multi-language support, and geographic expansion to regions like India, Canada, and Japan. Agentforce also now supports hosted LLMs from Anthropic and Google (upcoming), adding flexibility for regulated industries.

Salesforce’s commitment to trusted innovation is evident in new failover capabilities for resiliency, real-time streaming for responsiveness, and FedRAMP High Authorisation to extend Agentforce to public sector clients.

Industry-Ready and Built for Rapid Adoption

With more than 100 new prebuilt industry actions added, Agentforce 3 is positioned to deliver fast value across sectors, from patient scheduling in healthcare to campaign generation in advertising. Salesforce has also introduced flexible SKUs and usage-based pricing for its Service, Sales, and Industry Cloud editions, removing barriers to AI adoption.

Backed by global partners like Accenture, PwC, Deloitte Digital, and NeuraFlash, with over 270,000 certified specialists, Salesforce ensures enterprises can deploy, manage, and optimise their AI agent workforce with speed and assurance.

“Agentforce is helping PepsiCo unify insights and drive intelligent operations,” said Athina Kanioura, Chief Strategy and Transformation Officer, PepsiCo. “This is a pivotal step in our agentic journey.”

With Agentforce 3, Salesforce aims to make AI agent deployment not just scalable, but transformative.