Salesforce has launched Koa Decisions, a new model designed for agents that bypasses traditional large language model (LLM) text generation for faster decision-making. Instead of producing paragraphs, Koa Decisions takes a request and predefined categories to output a choice and a confidence score in milliseconds. This model, built on Google's Gemma and an evolution of Salesforce's HyperClassifier, aims to reduce the computational cost and latency associated with decision-making within agent workflows by providing a single-step output. AI
IMPACT Accelerates agent decision-making by replacing costly LLM calls with a faster, specialized model.
RANK_REASON This is a product launch for a specialized model that improves agent efficiency, rather than a core frontier model release.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →