:

SALESFORCE UNVEILS KOA, NVIDIA-POWERED CRM AI MODEL

AI DESK1 MIN READ
TUE, SEP 15, 2026

■ AI-SUMMARIZED FROM 1 SOURCE ▸ TIMELINE

Salesforce announced Koa, a specialized CRM reasoning model built on Nvidia's Nemotron architecture, at its Dreamforce conference. The model is designed to power AI agents within Salesforce's Agentforce platform.

Koa represents a collaboration between Salesforce and Nvidia to create a large language model optimized specifically for customer relationship management tasks. By leveraging Nvidia's Nemotron architecture, the model is engineered to handle CRM-specific reasoning and decision-making. The model will serve as the foundation for Salesforce's Agentforce, the company's AI agent platform that automates business processes and customer interactions. This specialized approach contrasts with general-purpose language models, focusing instead on domain-specific capabilities needed for sales, service, and marketing workflows. The partnership reflects broader industry trends of enterprises developing custom AI models tailored to their platforms and use cases. Salesforce's move positions Koa as a competitive alternative to general-purpose models for CRM-centric applications.

■ SOURCES

Techmeme

■ SUMMARY WRITTEN BY AI FROM THE LINKS ABOVE

■ MORE FROM THE AI DESK

Autonomous AI agents are actively harming internet infrastructure and content ecosystems. The widespread deployment of these systems is creating cascading problems across platforms and services.

JUST NOWAI Desk

Despite policies promising not to use corporate data for training, AI companies like OpenAI and Anthropic still face skepticism from major clients. When Anthropic announced it would retain usage logs for 30 days, Palantir, Nvidia, and Booz Allen Hamilton restricted the service from sensitive work.

JUST NOWAI Desk

U.S. Attorney General Todd Blanche has signaled a preference for limited enforcement action against AI companies and executives, opposing what he calls 'regulation by prosecution' as debate intensifies over the technology's risks.

1H AGOAI Desk

Bloomberg journalists will answer reader questions about the intensifying debate over whether AI development should be slowed, with the discussion taking place today at 2:30 p.m. EDT ahead of the US midterm elections.

1H AGOAI Desk

■ SUBSCRIBE TO THE DAILY BRIEF

ONE EMAIL, 5 STORIES, 06:00 UTC. UNSUBSCRIBE ANYTIME.