Salesforce Koa Decisions turns text, images, and audio into single-step decisions
Salesforce introduced Koa Decisions, the next version of HyperClassifier, designed to turn text, image, and audio inputs into classification, routing, permission, and scoring decisions in milliseconds.
AT A GLANCE
Key points
- Koa Decisions returns a category, binary choice, or score in one step.
- The model supports text, image, and audio inputs.
- Salesforce plans integrations with Flow, Apex, prompt templates, and an API.
What is changing?
Salesforce announced Koa Decisions, the next version of the HyperClassifier model used to route requests inside Agentforce. The model reads a request together with a predefined list of options and returns a category, yes-or-no choice, or score in a single step.
Salesforce says Koa Decisions was developed from Google’s open-weight Gemma model using new training methods and adds image and audio support alongside text. In Salesforce’s own routing measurement, the decision model took about 26 milliseconds compared with roughly 1.4 seconds for a general-purpose large language model.
Salesforce plans to extend Koa Decisions beyond Agentforce routing into Flow, Apex, prompt templates, and an API for developers. The announcement does not mean all of those surfaces are generally available today.
Explore the original sources
Source date: 6 November 2026.
Published by AI Radar: 6 Nov 2026, 23:19 (Istanbul, UTC+3).
FOLLOW THE TOPIC
Topics and tags
Share your discovery.
Send this story to a friend.
Reader comments
What do you think about this development? Share your experience or question. Comments are published after editorial approval.
Your name/nickname and comment become public if approved. Read: Privacy notice.
Loading comments…