Introduction to Pi
Pi is a coding agent you talk to in your terminal.
$ open "Run your first pi session"
Meet Pi, a minimal terminal coding harness you can reshape. Launch a real pi session and have your first conversation about the starter p...
checkpoints run as you work
✓ checkpoint passed, lab 2 unlocked
What you'll build
Curriculum
9 labs, about 1 h
- 1
Run your first pi session
Meet Pi, a minimal terminal coding harness you can reshape. Launch a real pi session and have your first conversation about the starter project.
6 min
Pro - 2
Watch pi use its tools
A model writes text. Tools are what let it act. Send pi after a real bug in the starter project and watch it search, read, and edit.
8 min
Pro - 3
Show pi what you see
Bad answers usually mean bad context. Run a command with ! so pi reads the real output, and point at files with @ instead of describing them.
7 min
Pro - 4
Steer the work while it runs
You do not have to wait for a bad run to finish. Send a correction mid task, watch it land between steps, and stop a run cleanly when you need to.
8 min
Pro - 5
Pick up where you left off
Every pi conversation is saved to a file as you go. Name one, do some work in it, find where it lives, and learn how to reopen it tomorrow.
7 min
Pro - 6
Give pi your project rules
Stop repeating your conventions in every message. Write an AGENTS.md once and pi reads it on every turn from then on.
8 min
Pro - 7
Make your own slash command
Turn a request you make constantly into a command of your own. Write a prompt file, give it an argument, and run it on a real file.
9 min
Pro - 8
Teach pi a procedure
Hand the agent a step by step procedure it can follow every time. Write a skill in plain markdown, run it, and read what it produced.
9 min
Pro - 9
Put it to work
Bring it together. Build a real feature and watch your rules, your command, and your skill do the work you set them up for.
10 min
Pro
Your instructor
Elvis Saravia
Founder, DAIR.AI
Elvis founded DAIR.AI and wrote the Prompt Engineering Guide, one of the most widely used references on working with language models. He has worked at companies like Meta AI and Elastic, and specializes in AI agents, RAG, and harness engineering.
How labs work
Live workspace
You run the real tools in a browser workspace. Nothing to install.
Checkpoints
Each lab checks your files and output, then unlocks the next one.
Your work stays
The workspace persists between labs so you finish with a project.