LangChain from the ground up

Chains, prompts, parsers and tools

Intermediate
10 chapters
153 min read

What LangChain actually does, written out once in plain Python: chains as steps, prompts as templates, messages as the model’s only memory, parsers as the boundary where text becomes data, and tools as a validated dispatch table. Every chapter has a runnable example beside the framework’s own API.

What You'll Learn

Explain what a chain is without referring to a framework
Build and test a prompt template with named slots
Send a message list to a chat model and read the reply object
Parse model output into validated data, with a failure path
Compose steps with pipes, including parallel branches
Describe a tool to a model, validate its request, and execute it
Keep conversation history within a token budget
Assemble a small chain and a bounded agent loop from scratch

Prerequisites

Comfortable with Python functions, dictionaries and classes
No prior experience with LangChain or any LLM framework
The runnable examples use nothing outside the standard library

Vocabulary

The terms the chapters assume, in plain words.

callback: A function the library calls at fixed points in a run, such as when a step starts or ends.
chain: A series of steps where each one passes its result to the next, ending in the answer.
invoke: The method that runs a step once with one input and returns its result.
memory: Stored history from earlier turns, added back to a request so the model can refer to what was said.
message: One labelled turn in a chat request, marked as coming from the system, the user or the model.
model: The trained program that reads a message list and returns the next turn.
output parser: The step that turns a model's raw text into a value your program can use, or raises an error.
pipe: The operator that joins two steps into one new step, running the left side first.
prompt: The text a model receives for one request, built from fixed wording and the values filled into it.
prompt template: A reusable prompt with named blanks that are filled in for each request.
runnable: Any step with a standard way to run it, so steps can be joined without knowing each other's code.
streaming: Sending a reply in pieces as it is produced, rather than waiting for the whole text.
tool: A function exposed to a model, with a name, described inputs and a result.
tool call: A model's request to run a named tool with specific arguments, which your code may accept or refuse.

Course Content

01
What a chain actually is
Strip the branding away and a chain is a list of steps where each one hands its output to the next. Build one from plain functions before any framework is involved.
Beginner · 13 min read
Not started
02
Prompts are templates
A prompt with blanks in it is a template: fixed instructions, variable slots, and a rule about what happens when a slot is empty.
Beginner · 14 min read
Not started
03
Messages, not strings
A chat model takes a list of role-tagged messages and returns one. Understanding that list — and that the model remembers nothing between calls — explains most surprising behaviour.
Beginner · 14 min read
Not started
04
Output parsers
A model returns text; your program needs data. A parser is the boundary where that text becomes a value you can trust — or an error you can handle.
Intermediate · 15 min read
Not started
05
Composing with pipes
The pipe operator is one method: a step that knows how to call another step. Implement it once and the rest of the framework becomes obvious.
Intermediate · 15 min read
Not started
06
Tools and function calling
A model cannot run code. It can only ask for a function to be run, by name and with arguments — and your program decides whether to obey.
Intermediate · 16 min read
Not started
07
Memory and history
The model forgets everything between calls, so memory is something you build: a list you keep, a rule for what to drop, and a summary for what cannot fit.
Intermediate · 15 min read
Not started
08
Callbacks and tracing
A chain that cannot be observed cannot be debugged. Callbacks announce each run, and a trace turns those announcements into a tree with timings, tokens and errors.
Intermediate · 15 min read
Not started
09
Streaming and async
Streaming moves the first token earlier; async lets one worker wait on several calls at once. Neither makes the model faster, and both change how a chain is written.
Advanced · 16 min read
Not started
10
Capstone: a mini LangChain
Assemble the pieces the course has built — templates, a model step, parsing, a tool loop — into one small library, then compare it with what the real framework adds.
Advanced · 20 min read
Not started