The AI industry’s most visible breakthroughs have centered on consumer applications — personal assistants that write emails, generate images, and answer questions. These capabilities dominate benchmarks, capture headlines, and attract research investment. But the hardest challenges in enterprise AI live at the system level: autonomous agents that communicate across organizational boundaries without compromising security. For example, voice interfaces that handle the nuance of conversations. AI that understands context across evolving workflows even as the interface itself transforms. Systems that dynamically discover and optimize for emerging patterns.

To tackle these challenges, today Salesforce AI Research announced AI Foundry, an initiative focused on developing transformative AI innovation for the enterprise. AI Foundry brings together AI research, strategic customers, and academic partners to develop, test, and validate new AI capabilities designed to move from foundational research to product innovation faster than ever before.

Areas of Investment

AI Foundry is focused on strategic areas where Salesforce AI Research sees the greatest opportunity for enterprise impact:

• Simulation Environments: Agents that learn from experience
  Enterprise AI agents need to be tested against realistic business conditions before they reach production. AI Foundry has already developed eVerse, a  simulation environment  that exposes agents to thousands of edge cases, handoffs, and judgment calls, demonstrating how it is able to increase both capability and consistency in enterprise use cases. Recently, eVerse was used to stress-test Agentforce Voice across thousands of simulated conversations and to pilot UCSF Health’s contact center billing agents.

• Ambient Intelligence: Agents that disappear into the environment
  Ambient intelligence is context-aware, proactive, and timely. It understands the full situation, anticipates needs before they emerge, and surfaces just-in-time insights. AI Foundry is developing capabilities that embed this intelligence directly into enterprise workflows, with a deliberate focus on human-AI interaction patterns. The goal is AI that’s always on but never overwhelming, helpful without creating information overload. 

• Agent-to-Agent Ecosystems: Agents that interact on behalf of companies across organizational boundaries
  AI Foundry has developed protocols, such as agent cards, and is continuing to invest in building an enterprise multi-agent semantic layer, with standardized protocols, guardrails, decision logging, and coordinated escalation. Working with Salesforce’s legal council, as well as with the Office of Ethical Use of Technology, our research team is defining the legal frameworks required for autonomous agent negotiation.

Research to Product, Faster

AI Foundry brings together foundational research and human-centric design principles  to drive real innovation into Salesforce’s product roadmap much faster than the traditional product cycle would allow. “Many of the old rulebooks simply don’t apply anymore,” said Itai Asseo, VP of Salesforce AI Research. “AI Foundry connects foundational research to real business problems by collaborating closely with our strategic customers in rapid iteration cycles.”

The initiative also extends Salesforce’s academic grant program and partner ecosystem, connecting external researchers and cross-industry talent to partner on solving the most pressing enterprise AI challenges.

Why This Matters Now

Enterprise leaders recognize what the benchmarks don’t capture: Business AI operates under requirements that consumer applications never face. Only fit-for-purpose business systems can handle the complexity, trust, and operational reality that enterprises demand. 

AI Foundry is designed to tackle these challenges — not by chasing model benchmarks but by building the systems, protocols, and validation infrastructure that make AI agents work for business at scale. 

Learn more:

• Learn about Salesforce AI Research at SalesforceAIResearch.com
• Read about eVerse, the enterprise simulation environment for voice and text agents
• Explore Chief Scientist Silvio Savarese’s take on agent-to-agent ecosystems