For Beginners¶
You know Python. You've heard about LLMs. You want to build something — but you don't know where to start.
This section is for you.
No prior experience with LLM APIs, agent frameworks, or "AI infrastructure" is required. By the end you'll have built a working multi-agent pipeline from scratch, and you'll understand why every piece exists.
What you'll learn¶
Step 1 → What is an LLM and how does an API call actually work?
Step 2 → What does raw SDK code look like? (OpenAI / Anthropic / Gemini)
Step 3 → Your first LazyBridge agent — every line explained
Step 4 → Giving your agent tools (so it can actually do things)
Step 5 → Why multi-agent + sub-agent as a tool (the flexible pattern)
Step 6 → Cross-model verification with verify= (let Claude judge GPT)
Step 7 → Sequential pipelines with Agent.chain
Step 8 → Parallel work with Agent.parallel
Step 9 → Explicit DAGs with Plan & sentinels
Step 10 → Routing — conditional branching on outputs
Step 11 → Human in the loop with HumanEngine
Step 12 → LazyBridge vs LangGraph vs CrewAI — which one for what
Step 13 → Where to go next
Each step builds on the previous one. Skip ahead if you already know a topic, but the examples in later steps assume you've seen the earlier ones.
Prerequisites¶
- Python 3.10+ installed and working
- You can write a function, use a
forloop, and read a stack trace - That's it
You do not need to know: machine learning, transformers, embeddings, vector databases, async programming, or any other AI/ML concept. We'll introduce everything as we need it.
What LazyBridge is (one paragraph)¶
LazyBridge is a Python library that lets you build LLM-powered agents with the minimum amount of code. An agent is a program that uses an LLM to decide what to do — it can call functions you write, talk to other agents, remember context across turns, and return structured results. LazyBridge handles all the plumbing (API calls, retries, tool dispatch, result parsing) so you write business logic, not boilerplate.
What you'll build by the end¶
A pipeline that takes a research question, searches the web, summarises the findings, and returns a one-paragraph report — using three specialised agents that hand off work to each other automatically.
API keys
The examples in this tutorial use real LLM providers (Anthropic, OpenAI, Google). You'll need at least one API key. If you don't have one yet, sign up for the Anthropic Console — they offer free credits for new accounts, and LazyBridge defaults to Anthropic.
Every example also has a mock variant that runs without any API key, so you can follow along even before you have credentials.