Salesforce today launched Agentforce 3, a major update to its flagship Agentforce platform designed to help enterprises scale AI agents with greater transparency, interoperability, and control. Unveiled at a time of rising enterprise demand for “digital labor,” the release introduces a centralized Command Center, open integration protocols, and a suite of new tools to monitor, manage, and optimize AI-driven workflows.

At the heart of Agentforce 3 is the Command Center, a unified observability hub that gives enterprises real-time insights into AI behavior. Through dashboards and alerts, teams can monitor agent health, track latency and error rates, and drill into individual interactions—enhancing trust and accountability as digital agents take on more autonomous roles. The Command Center also offers AI-powered recommendations, helping users refine agent performance proactively.

Interoperability takes a leap forward with native support for the Model Context Protocol, enabling plug‑and‑play compatibility with external AI models and tools. This simplifies integrations across ecosystems without custom coding, ensuring secure and governed collaboration between agents from multiple vendors.

Furthermore, Agentforce 3 expands its reach with over 100 prebuilt industry actions, fast‑tracking deployments in sectors such as retail, financial services, healthcare, and beyond. It also introduces web-search grounding and inline citations, enabling agents to draw on verified sources and reduce hallucinations—boosting accuracy for customer interactions or internal use cases.

Performance and resilience have been enhanced with model failover capabilities and faster response streaming, delivering up to 50 percent lower latency and smoother conversational experiences. Localization is another focal point: the platform is now available across regions including Canada, the UK, India, Japan, and Brazil, with support for multiple languages and plans for further globalization.

Architecturally, Agentforce continues to refine its Atlas Reasoning Engine and integration with Data Cloud, powering scalable inference pipelines capable of processing billions of predictions monthly. This enables agents to retrieve real-time data, apply business logic, and execute complex workflows—ranging from auto-engaging leads to resolving support cases—while ensuring data fidelity and governance.

Salesforce leadership emphasizes that these upgrades address the key challenges of scaling AI agents: observability, operational control, integration flexibility, and global readiness. Agentforce 3 reflects the maturity of agentic AI, which some analysts are heralding as the “third wave” following generative and predictive AI.

More than 8,000 organizations have deployed Agentforce, with usage soaring—internal metrics show a 233 percent growth in agent activity over the past six months. Early adopters such as PepsiCo and Heathrow Airport already report notable impacts: automated responses, cost reductions, and improved speed in customer-facing and back‑office workflows.

Agentforce 3 marks a pivotal step in Salesforce’s AI journey. It reframes AI agents from isolated pilots into enterprise-scale autonomous collaborators that are transparent, interconnected, and reliable. By embedding observability, interoperability, and performance into the platform, Salesforce aims to empower businesses to harness agentic AI safely, broadly, and at scale—bridging human ingenuity with machine efficiency.