- ホーム
- > 洋書
- > 英文書
- > Computer / General
Full Description
Build faster, safer, and more scalable AI agents with decision models like Jev and open-weight alternatives. Apply fast, structured decisions to classification, scoring, routing, agentic gating, and agent workflows.
Key Features
Replace slow generative LLM calls with decision models like Jev that respond in milliseconds
Run Jev and open-weight decision models with LangChain, hosted or locally with Ollama
Gate agent actions with confidence thresholds to improve agent security, governance, and control
Book DescriptionGenerative LLMs excel at reasoning and open-ended tasks, but not every decision in an AI system requires generation. Agents frequently need fast, structured decisions about tool selection, classification, routing, and whether an action is safe to execute.
This practical guide introduces System One decision models like Jev, which return probabilities, scores, and classifications instead of free-form text. You'll learn to combine these models with generative LLMs to build faster, more efficient agentic systems.
Through hands-on examples, you'll integrate Jev and open-weight alternatives with LangChain, run models through hosted APIs or locally with Ollama, and build classification, scoring, triage, and routing workflows.
You'll also explore agentic gating as a security and governance layer. Using confidence thresholds and policy-driven controls, you'll determine when agents can execute tools, access sensitive capabilities, request human approval, or escalate decisions.
By the end, you'll know how to reduce unnecessary LLM calls, improve latency and cost efficiency, and build scalable AI agents with greater security, reliability, and control.What you will learn
Control tool execution using confidence thresholds
Implement agentic gating for security and governance
Grade responses and bugs using continuous scores
Build triage and evaluation workflows
Run open-weight models locally using Ollama
Integrate decision models with LangChain workflows
Use Jev for fast classification, scoring, and routing
Design hybrid System One and System Two agent architectures
Who this book is forThis book is for AI engineers, backend developers, and software developers building AI agents and agentic systems that require fast, structured decision-making. It is particularly useful for developers looking to complement generative LLMs with decision models like Jev for classification, scoring, routing, agentic gating, security, governance, and control. Basic programming knowledge and familiarity with APIs, JSON, and AI concepts are recommended.
Contents
Table of Contents
Understanding Jev and Why Structured Decision Models Matter
Modeling Decisions with Jev's System One Primitives
Integrating Jev with LangChain
Building Cost-Efficient Agents with Confidence Gating and Model Routing



