Salesforce Built a Cheaper AI Helper for Sales and Support

Salesforce announced Koa at its Dreamforce tech conference, its first reasoning model. It was built with Nvidia and will be available in Agentforce as another choice alongside the other models Salesforce offers there. The announcement was reported Sept. 15, 2026. TechCrunch

Koa is built on an Nvidia model called Nemotron, which is open-weight. That means its basic settings can be downloaded and reused by others. Salesforce and Nvidia did further training together to handle sales, marketing and customer-support tasks. That scope is narrow by design. It targets the workflows Agentforce already serves rather than general conversation.

Agentforce is Salesforce's platform where customers build agents to handle tasks like answering customer service questions or scheduling appointments. Before Koa, long-running or multi-step reasoning requests entering through its AI gateway were routed to frontier models like Claude or ChatGPT. Koa gives customers another option inside that same gateway.

The joint work connects Salesforce AI, where Jayesh Govindarajan is EVP, with Nvidia's enterprise generative AI software organization, where Kari Ann Briski is VP of Generative AI Software for Enterprise. The technical approach relied on synthetic data mimicking customer patterns, not actual Salesforce customer data.

Salesforce positions Koa as an open-weight alternative to closed frontier models. The claim is token efficiency. It uses fewer tokens for the same work. Tokens are the small pieces of text a model reads and writes. Token budgets matter. For agentic runs that chain tool calls, retrieval steps and validation passes, inference cost scales directly with tokens consumed.

Looking at what this means for teams running agents in production, the shift is practical. A gateway that can keep routine work in a focused model and call on a larger model only when needed changes cost and control. It also changes debugging. Records from a smaller model built for sales work are easier to check than records from a general system asked to do that job.

In my view, the synthetic-data detail deserves as much attention as the token claim. Training on data that mimics customer patterns, rather than on customer records themselves, addresses a constraint every enterprise AI team recognizes. Access to production data is limited by contract, regulation and trust. If mimicry closes enough of the gap for sales and support dialogue, that pattern will spread. It enables iteration without moving sensitive records into training.

The broader context here is evaluation. The practical test will be straightforward. Does Koa hold context across multi-step service flows. Does it call tools reliably. Does it stay within token budgets under load. Open weights help here, because evaluation can move beyond vendor demos into repeatable tests. That gives enterprise teams something they can measure, version and keep.

Looking further out, the longer arc favors specialization, like using a family doctor for routine visits and a specialist only when needed. General frontier systems will continue to handle novel reasoning, while tuned models absorb repetitive commercial work. If Koa performs as described, Agentforce users gain leverage without changing how they build agents.