Salesforce today announced the launch of AI Foundry, a new internal service aimed at accelerating agentic AI research and development. The initiative is designed to bridge the gap between foundational research and enterprise-grade AI products, enabling faster innovation cycles and more impactful deployments. The company emphasized that AI Foundry will collaborate closely with strategic customers and academic partners to co‑develop and test new AI tools and capabilities.(itpro.com)
A key feature of AI Foundry is eVerse, a simulated environment that exposes AI agents to thousands of edge cases, handoffs, and judgment calls—allowing rigorous stress‑testing before deployment. This capability has already been used to evaluate Agentforce Voice through thousands of simulated conversations.(itpro.com)
Another focus area is ambient intelligence—agents that operate context‑aware and proactively within enterprise workflows, effectively “disappearing into the environment.” Salesforce is experimenting with these always‑on agents to support workers on the move by anticipating needs and delivering just‑in‑time insights.(itpro.com)
Salesforce is also exploring agent‑to‑agent interactions, including cross‑organizational ecosystems. The company is developing protocols such as “agent cards” to help enterprises verify third‑party agents’ legitimacy and purpose. Legal frameworks for autonomous agent negotiation are being defined in collaboration with Salesforce’s legal counsel and its Office of Ethical Use of Technology.(itpro.com)
This move underscores Salesforce’s commitment to advancing enterprise AI from static automation toward dynamic, autonomous agent ecosystems. By embedding research directly into product development and emphasizing governance, simulation, and ethical frameworks, Salesforce aims to deliver AI that is both powerful and trustworthy.