Be10x · Daily Drop

Stop scripting it.
Let it decide.

Five drops on building intelligent agents with LangChain. One question, three tool calls, and a model that chose how many times to go round — written in about twenty lines of Python.

The loop you're building
Your question Model decides each turn tool calls? lookup_order() search_policy() days_since() ToolMessage goes back to the model no tool calls Final answer

Three tools, three trips, one answer — and nobody wrote the number three. That is the whole difference between a chain and an agent, and it is what create_agent gives you.

MonPDF
One question, three tool calls
A 12-slide deck, cover to close. Swipe through it, or hit Save as PDF.
TueHands-on
Assemble the agent
Build a working agent line by line, then fix six snippets of deprecated code.
WedInteractive
Stop scripting it. Let it decide.
Six live widgets — rewrite a tool's docstring and watch the model change its mind.
ThuFlash cards
Twelve ideas behind a LangChain agent
create_agent, @tool, docstrings, messages, checkpointers. Tap to flip.
FriQuiz
Ten questions on agents in LangChain
Current API and stale-tutorial spotting, with an explanation after every answer.

All five teach LangChain v1, where create_agent replaced four older ways of building an agent. If you have learned from blog posts before, start with Monday's deck — a good deal of what is still online was superseded in October 2025.