How to build AI Agents with Claude API in Python

Name: How to build AI Agents with Claude API in Python
Rating: 4.4 (890 reviews)

Learn AI Fundamentals with Claude API, how to write good prompts and evaluations, RAG, MCP, Vector Databases

Role Play

Created byJustin Barnett

Last updated 5/2026

English

What you'll learn

What are AI Agents and Retrieval Augmented Generation (RAG)
How to create a RAG pipeline from scratch
Claude 3 API Overview, differences between the 3 models, pricing, and rate limits
How to write good prompts and evals to iterate on AI applications

Course content

5 sections • 30 lectures • 2h 38m total length

Introduction1:04
Explore ai fundamentals, llms and transformers, and learn prompt writing, evals, rag retrieval, augmented generation, and few-shot prompting, plus cloud api concepts, model differences, and rag reranking.
Familiarize yourself with the Udemy User Interface2:41
Familiarize yourself with the Udemy user interface to maximize your learning, including navigating the course homepage, accessing downloadable PDFs, and using the Q&A, notes, announcements, and learning tools.
Connect with Me0:53
Join a private Facebook group called ChatGPT and Gen I for content for enrolled students to connect with instructors and classmates, ask questions, and discuss gen AI topics.
AI Definitions and Acronyms5:25
Learn essential ai definitions and acronyms, including large language models, neural networks, generative adversarial networks, transformers, and api-driven ai agents, plus subareas like machine learning and computer vision.
Testing AI Definitions and Acronyms

What are LLMs (Large Language Models)3:30
Discover how large language models infer the next word from probabilities and why they are not knowledge models. Harness their power with prompt engineering and detailed inputs.
How do Transformers Work?10:12
Discover how transformers use attention to weight training data, encode sentences with n-grams, and form inference from prompts, while non-determinism drives hallucinations that require careful fact-checking.
Basics of Prompt Engineering5:59
Explore the basics of prompt engineering for large language models, covering text summarization, information extraction, question answering, code generation, and effective prompting techniques for robust AI apps.
Zero Shot and Few Shot Prompting3:55
Understand zero shot and few shot prompting, and how prompts use pre-built knowledge or provided examples. Recognize their limitations in complex reasoning and numeric tasks.
How to Evaluate Prompts and Models5:32
learn to evaluate prompts and models using rubrics, inputs, and scores, and iterate prompts with empirical feedback. explore multiple choice, exact match, and rubric-based open-ended evaluations to improve prompt performance.
A Framework to Writing Good Prompts5:18
This lecture introduces a ten-part prompt framework—context, tone, background data, details, task rules, examples, history, immediate tasks, thinking step by step, output format, and edge cases—plus testing for production readiness.
What makes a Model Multimodal?7:18
Explore multimodal lms that combine text with image, audio, and video tokens via alignment and modality modules, enabling richer inputs and improved accuracy.
What is RAG (Retrieval Augmented Generation)6:41
Master retrieval augmented generation (rag) to boost llm accuracy by sourcing internal and external data, citing sources, and delivering contextually relevant answers with improved trust.
What is MCP (Model Context Protocol)8:29
Explore the model context protocol (mcp), an open standard for seamless integration of tools, data sources, and prompts within a client–server architecture for ai agents.
What is Fine Tuning5:34
Learn how to fine tune pre-trained LLMs with a fine tuning set, using OpenAI GPT-3.5, and identify practical use cases and when not to fine tune.
Review LLM Concepts

Model Differences and Pricing: Opus, Haiku, Sonnet2:39
Compare haiku, sonnet, and opus cloud models to balance speed, cost, and performance; haiku is fast, sonnet balanced for enterprises, opus for tasks. Pricing relies on input and output tokens.
API Capabilities3:19
Explore cloud capabilities for Claude 3 API in Python, from dialogue and summarization to code tasks and content generation, including multilingual, multimodal retrieval and tools integration via JSON schema.
Getting an Anthropic API Key1:30
Create an anthropic api key in the dashboard and copy the secret. Set up billing with a phone number and a credit card on pay-as-you-go plan with $5 free credit.
Client Libraries (SDKs)1:46
Explore the main client libraries for the Claude 3 API, including Python's anthropic package and TypeScript SDK, creating an anthropic object, configuring model, max tokens, and messages.
Using the Messages API5:48
Use the messages API to send prompts, specify models and tokens, and stream responses; configure top k, top p, and temperature, and learn to manage tools and cloud integrations.

Strategies to Reduce Latency3:35
Choose the right model for latency among haiku, sonnet, and opus. Consider baseline latency and time to first token as key metrics, and use streaming to improve perceived speed.
Rate Limits2:44
Explore rate limits for the Claude 3 API, including request per minute, tokens per minute and per day across free, build, and scale tiers; compare haiku, sonnet, and opus models.
AI Application Architecture8:20
Navigate the retrieval augmented generation architecture for AI applications. See how a datastore feeds embeddings into a vector database for fast semantic retrieval and LLM-driven responses.
Building a RAG Pipeline from Scratch14:59
Build a retrieval augmented generation pipeline from scratch by chunking text, creating embeddings, indexing with a vector store for semantic search, and querying an LLM with retrieved context and citations.
RAG Reranking for Limited Context Windows10:09
Learn to use a vector database (Lance) to perform reranking for limited context windows, combining embedding search with a second-pass reranker to select top k contexts for the language model.
What are AI Agents9:20
Define AI agents as software that interact with an environment to achieve goals using multiple models, a vector store for retrieval augmented generation, and coordinated execution and task prioritization workflows.
Open Sourced Models vs Closed Source Models10:40
Compare closed source and open source llms, weighing the pros and cons of self-hosting versus managed hosting, and learn how cost, control, and maintenance influence model choice.

Requirements

Basic understanding of Python
Some programming experience is helpful

Description

Step into the forefront of AI Agent development by mastering Anthropic's groundbreaking Claude 4 model suite! If you're a developer eager to leverage state-of-the-art Large Language Models, this course provides the essential knowledge and practical skills to integrate Claude 4 Opus, Sonnet, and Haiku effectively into your applications. Go beyond basic API calls and learn to architect sophisticated, reliable AI-powered features.

Here's what you'll master in this comprehensive course:

Understand Core AI Concepts: Grasp the fundamentals of LLMs (Large Language Models), Transformers, Multimodality, RAG (Retrieval Augmented Generation), MCP (Model Context Protocol), and Fine-Tuning principles essential for effective API usage.
Engineer Prompts for Agents: Develop advanced Prompt Engineering skills specifically tailored for guiding agent behavior, planning, and tool usage via the API.
Master Prompt Engineering for APIs: Learn Zero-Shot/Few-Shot Prompting, evaluation frameworks, and techniques to achieve precise, desired results when interacting programmatically with the Claude 3 API.
Dive Deep into the Claude 3 API: Compare Opus, Sonnet, and Haiku models (capabilities, strengths like reasoning/speed/cost, pricing), manage Anthropic API keys, utilize official Client Libraries (SDKs), and expertly use the versatile Messages API for complex interactions.
Design Robust AI Application Architecture: Implement practical strategies to mitigate latency, effectively manage rate limits, plan for scaling your AI features, and understand architectural patterns like AI Agents.
Build and Optimize RAG Pipelines: Gain invaluable hands-on experience constructing a Retrieval Augmented Generation (RAG) pipeline from scratch and applying advanced reranking techniques to enhance context utilization and grounding.
Explore Agent Design Patterns: Understand the differences between Open Source vs Closed Source Models in the context of agent development and explore common architectural patterns.

Building with powerful APIs requires smart architecture. This course tackles real-world developer challenges head-on, moving beyond simple API calls to focus on creating resilient and scalable systems. You'll learn not just how to use the Claude 3 API, but how to integrate it thoughtfully within a larger application context.

This course transforms you from merely calling an API to architecting intelligent, robust, and scalable AI solutions. Whether you're building advanced chatbots, sophisticated content generation tools, complex analysis engines, or entirely new AI-powered products, you'll gain the confidence and expertise to leverage the full potential of Anthropic's cutting-edge Claude 3 models.

Enroll today and start building the future with the Claude 3 API!

Who this course is for:

Beginner Python developers learning more about building AI Applications
Business owners that want to understand how AI will transform their business
Project managers that want to learn more about how AI application architecture work

How to build AI Agents with Claude API in Python

What you'll learn

Explore related topics

Course content

Introduction4 lectures • 10min

AI Overview11 lectures • 1hr 2min

Claude 3 API Overview5 lectures • 15min

AI Application Architecture7 lectures • 1hr

Tool Calling with Claude4 lectures • 11min

Requirements

Description

Who this course is for: