Silvio Savarese on operational intelligence: the AI layer that makes enterprise signal legible, so leaders can act before the moment passes.

5 requirements for trustworthy AI agent-to-agent communication, from Salesforce's Chief Scientist and Chief Legal Officer. Standards, identity, and accountability, explained.

Today, Salesforce is expanding its AI model cards with standardized environmental impact metrics. This update helps customers better understand the energy

Post-training methods (RLVR, On-policy distillation) are Episode-local Language models are getting better at learning from feedback during post-training. In reinforcement learning with verifiable rewa...

Negotiation requires more than inferring what the other side wants: it requires using that information to make advantageous offers and counteroffers over multiple turns. We study whether large languag...

A perspective on the future of enterprise AI that is ambient: context-aware, always on, and invisible to users.

What if launching an RL training run felt less like operating a GPU cluster and more like talking to a sharp research engineer in Slack? Training AI models is still strangely artisanal, involving…

Enterprise deep research often fails to produce decision-ready reports due to uneven information coverage, context explosion, and premature stopping. We propose a scalable Enterprise Deep Research (ED...

Recap from the Salesforce TDX 2026: AI Agents, 3 key AI trends and how engineers use AI.

Salesforce AI Research will present 21 accepted papers at ICLR 2026, the Fourteenth International Conference on Learning Representations. The conference runs April 23–27 at the Riocentro Convention an...

As codebases grow to millions of lines of code, can AI agents still understand, reason, and code effectively? LoCoBench-Agent delivers the answer: a comprehensive benchmark for evaluating AI coding as...

On-the-ground reporting from the must-attend developer event for the Agentic Enterprise

As AI matures, enterprise success hinges on integrated systems that deliver consistent performance across the most complex professional business workflows

Details on Salesforce AI Foundry, launched today to help AI researchers, customers and partners to collaborate in a safe environment

The era of the model wars is over, with enterprise AI success occurring at the system level, demanding reliability and full integration over model power

Salesforce showcases ambient intelligence for real-time sales workflows and highlights Agentforce upgrades focused on scale, governance, and enterprise outcomes.

Simulation environments, agent-to-agent ecosystems, and ambient intelligence will be at the heart of the Salesforce product roadmap through AI Foundry initiative.

Itai Asseo, VP of Salesforce AI Research, explains some new enterprise realities on the way.

Tired of GUI automation that breaks after one demo? Discover how GPA reimagines process automation to deliver stable, scalable, and truly reliable results for enterprise workflows.

During live sales calls, customers frequently ask detailed product questions that require representatives to manually search internal databases and CRM systems. This process typically takes 25-65 seco...

Explore how Salesforce AI Research is optimizing Reinforcement Learning (RL) for the agentic era. Discover new frameworks for building efficient, scalable AI agents that power the future of autonomous...

As frontier models mature and become commodities, innovation in enterprise AI is happening at the system level, not the model level.

As frontier models mature and become commodities, innovation in enterprise AI is happening at the system level, not the model level.

As frontier models mature and become commodities, innovation in enterprise AI is happening at the system level, not the model level.

VIBEPASS, a new benchmark, reveals a fundamental weakness in modern AI coding assistants: even with near-perfect scores on code generation tasks, frontier models falter when it comes to finding and fixing subtle bugs...

AI agents are rapidly advancing from passive language models to autonomous systems executing complex, multi-step tasks. Yet their overconfidence in failure remains a fundamental barrier to deployment ...