This page compiles 1 file from the repository, verbatim, in reading order. The living version: this folder on GitHub.

level-2-intermediate/README.md

Level 2 — Intermediate

You are: comfortable using AI daily. You will become: a builder — someone who can build agents, RAG systems, MCP servers, evals, and cost-controlled workflows. Typical time: 60–90 days at 4–8 hours/week.

Level 2

Prerequisites

Level 1 exit criteria met — see ../LEVEL-MAP.md
Comfortable in a terminal
Basic Python or TypeScript
Familiarity with git and GitHub
A codebase or dataset of your own to anchor examples on

Exit criteria

You've moved to Level 3 when: - You've built 3 real custom agents on your own stack - You've deployed at least one MCP server - You've built a RAG pipeline over your own docs - You have an eval harness with regression testing - You can articulate cost per workflow in dollars - For companies: eval framework live, observability platform running, first agents in production

Tracks

individual/ — 12 core modules + 1 bonus for personal capability
company/ — 8 modules for org-level capability
repos/ — 4 public GitHub repos backing the core walkthroughs — read the overview
shared/ — templates and briefs
eli10/ — plain-language companions for each big concept

Working code repositories

Repo	For walkthrough	Domain
`lease-abstractor`	`02-build-first-agent`	Real estate
`support-triage-agent`	`03-build-second-agent`	Any (SaaS, e-commerce, services, agency)
`diligence-agent`	`04-build-third-agent`	Any (M&A, vendor, CRE, contracts)
`belle-mcp-server`	`05-mcp-server`	Real estate

How to work Level 2

Pick one of the three agent walkthroughs (02, 03, 04) and build it end-to-end. Not skimmed. Built.
Then build an MCP server (05) exposing your own data.
Then build a RAG pipeline (06) over your own docs.
Then wrap the eval harness (07) around all of them.
Then optimize cost (08).
Everything else deepens what you've built.
Bonus: individual/13-combine-two-agents — chain two of your agents into a real system