L2: Active pairing
The AI writes alongside you all day, handling the boring parts while you steer and review everything as it lands. Shapiro estimates about 90% of AI-native developers live here.
The path.
- CLI agents Two families here: IDE assistants (autocomplete + chat + in-editor agents living in your editor) and CLI/terminal agents (you talk to an agent in the shell.
- AGENTS.md & CLAUDE.md In Context, Skills & Memory Agents are only as good as what they know when they start, and by default they start knowing nothing about your repo, your conventions, or last week's decisions.
- What an agent actually is In Agentic Engineering: Core Ideas "Agentic coding" is an overloaded term.
Check yourself.
One question per step. Take it cold to find where to start, or after reading to see what stuck. Nobody's grading you.
Best so far:
Question 1 of 3
-
What was the defining 2025→2026 shift in coding tools?
The shift was to agentic tools that plan, edit many files, run commands, and iterate, with pricing largely moving to usage credits on top of a seat.
Covered in step 1 · CLI agents -
Beyond build and test commands, what makes AGENTS.md most valuable?
AGENTS.md is the vessel for compound learning: hard-won lessons and "why we don't" decisions get written back so every future session and parallel agent starts smarter.
Covered in step 2 · AGENTS.md & CLAUDE.md -
A working coding agent fits in ~400 lines: "an LLM, a loop, and enough tokens." What follows from that?
If the loop itself is a few hundred lines of boilerplate, nearly everything that makes one agent better than another lives in the harness around the model. That's the section's key lesson.
Covered in step 3 · What an agent actually is
-
L0: Manual
The ground floor: AI as a search engine and occasional tab-complete. No real productivity change.
Everyone starts here. There's nothing to unlock yet.
-
L1: Discrete task offloading
You hand the AI small, self-contained jobs: write the unit tests, draft the docstring, generate one well-scoped function. The real work is still yours.
Chat assistants · Prompting 101 · IDE assistants
Explore path -
L2: Active pairing
The AI writes alongside you all day, handling the boring parts while you steer and review everything as it lands. Shapiro estimates about 90% of AI-native developers live here.
CLI agents · AGENTS.md & CLAUDE.md · What an agent actually is
You're on this rung
-
L3: Human-in-the-loop management
You mostly stop typing code. Agents run several tasks at once and your day becomes reviewing their work. Most people never go past this level.
Own the outer loop · Agentic code review · Role separation · MCP & tools
Explore path -
L4: Specification-driven
You define what to build instead of building it: write the spec, set the tests, judge the results. Leave for twelve hours and come back to a green suite.
The harness as an artifact · Evals-as-tests · LLMOps & guardrails
Explore path -
L5: The dark factory
Agents building the software with humans out of the loop. Shapiro reports it working only for teams under five people, and calls it likely our future.
Swarms & fleets · The factory model · Human factors
Explore path