A practical 
 guide to 
 building agents Contents What is an agent? 4 When should you build an agent? 5 Agent design foundations 7 Guardrails 24 Conclusion 32 2 Practical guide to building agents Introduction multimodality, and tool use have unlocked a new category of LLM-powered systems known as agents. This guide is designed for product and engineering teams exploring how to build their first agents, distilling and effectively. After reading this guide, you’ll have the foundational knowledge you need to confidently start building your first agent. 3 A practical guide to building agents What is an agent?

its findings, and create insightful multi-page, reports that you can turn into engaging podcast-style conversations… …It’s a step towards more agentic AI that can move beyond simple question- answering Performance = +225x Over Eight Years 106 1 GPT-MoE Inference Workload = A type of workload where a GPT-style model with a Mixture-of-Experts (MoE) architecture is used for inference (i.e., making predictions) for object perception, but for path planning and vehicle controls. We replaced 330,000 lines of C++ code with neural nets. It's really quite remarkable. So, as a side note, I think Tesla is probably

efficacy: the model you use, the model’s training data, the model configurations, your word-choice, style and tone, structure, and context all matter. Therefore, prompt engineering is an iterative process high-quality prompts that guide LLMs to produce accurate outputs. This process involves tinkering to find the best prompt, optimizing prompt length, and evaluating a prompt’s writing style and structure in relation System, contextual and role prompting System, contextual and role prompting are all techniques used to guide how LLMs generate text, but they focus on different aspects: • System prompting sets the overall

Domain expertise Fine-tuned models better understand your industry’s terminology, style, and context. Consistent tone and style For a retailer, that could mean every product description stays true to brand other agents to get things done. We’ll continue to report back from the front lines of AI to help guide your own thinking. Product Note: Operator Operator is an example of OpenAI’s agentic approach

基础设施：编程语言 LLM 的训练和应用通常使用多种编程语言，取决于任务的需求和团 队的偏好。 。它的广泛使用得 益于其简洁的语法、强大的库支持（如 ）和深度学习框架（如 ）。 此外， ，C++ 有时 用于优化计算密集型任务，而 Java 在企业环境中处理模型部署和系 统集成方面常见。JavaScript 适用于 Web 环境的 LLM 应用。 13 / 32 LLM 基础设施：编程语言 年是大语言模型 (LLM) 之年，Python 作为人工智能领域使用度最高的编程语言，在 2023 年到底有多火？ 从各种开发者报告、编程语言榜单来看。只要出现有关编程语言流行度的排名， ，而 Java、C/C++ 等 同样在 LLM 开发中发挥关键作用的语言紧随其后。 14 / 32 LLM 基础设施：编程语言  2023 年 9 月面向大众开放 创业公司 Modular AI 开 发  结合了

innovation, 
 e.g. use (GA/RL/BayesOpt/your favorite ML method) for AutoSchedule 
 Easy shift to C++ when product readyInterpolate with Other Compilers MLIR-TF Function relay::Function TorchScript

overview data weight1 weight3 weight2 output Build() Your graph representation (e.g., JSON, C++, etc) CompileExternalLib() Subgraph binary/library/engine (e.g., so, JSON, etc)© 2019, Amazon Web

Service 8 8 6.952 。 C++ RPC (Merged into Masten 6 4 2.353 2. ， 曾硬证 0 Mobilenet 1.0

interconnects.ai/p/deepseek-r1-recipe-for-o1 https://newsletter.maartengrootendorst.com/p/a-visual-guide-to-mixture-of- experts 教程作者：郭震，工作 8 年目前美国 AI 博士在读，公众号：郭震 AI，欢迎关注获取更多原创教程。资 料用心打磨且开源，是为了帮助更多人了解获取

methods not delivering generalized performance 2 Why TVM? XTVM for Speech Synthesis - WaveRNN-style model architecture - Autoregressive sampling net running at faster than real-time - Compute split