Salesforce AI Research has announced the launch of AI Foundry, a new initiative aimed at accelerating the development of enterprise-grade AI systems, shifting the focus from standalone models to integrated, system-level intelligence.
The move comes as large language models mature and become increasingly commoditised, prompting a shift in innovation towards how AI systems operate in real-world enterprise environments rather than how individual models perform in isolation.
AI Foundry brings together Salesforce’s research teams, strategic customers, and academic partners to develop, test, and validate AI capabilities that can move from research to production more rapidly. The initiative is designed to address the complexities of enterprise AI, including cross-system coordination, contextual understanding, and secure agent interactions.
According to Silvio Savarese, the key challenges in enterprise AI now lie beyond model performance, focusing instead on ensuring reliability, consistency, and scalability across interconnected systems.
AI Foundry will prioritise several strategic areas of investment. These include simulation environments such as eVerse, which enable AI agents to learn from complex, real-world scenarios before deployment; ambient intelligence capabilities that embed context-aware AI directly into workflows; and agent-to-agent ecosystems that allow AI systems to interact securely across organisational boundaries.
The initiative also aims to accelerate the transition from research to product innovation through closer collaboration with enterprise customers and rapid iteration cycles. Itai Asseo noted that traditional product development approaches are no longer sufficient for the pace of AI evolution, necessitating tighter integration between research and real-world use cases.
As enterprises increasingly adopt AI, the emphasis is shifting towards building systems that can handle operational complexity, maintain trust, and deliver consistent outcomes at scale. AI Foundry is positioned to address these requirements by focusing on infrastructure, protocols, and validation frameworks rather than model-centric benchmarks.
With this launch, Salesforce is reinforcing its strategy to enable enterprises to move beyond experimentation towards deploying reliable, system-level AI that can operate effectively within complex business environments.