Context-Aware Collaborator

The key conceptual jump. Claude knows who you are, what you're working on, and what you've already established.

If this is you

You've given Claude persistent context — a CLAUDE.md file, a Project description, or an equivalent
You work on actual files and documents, not just pasted text
Claude responds to your project's logic, not just to what you typed today

Your risk

⚠ Agenda drift and centroid capture

The AI now knows your project well enough to influence it. Its suggestions will be coherent with your framing — which is precisely what makes them dangerous. The centroid problem (Claude's picture of any field is its statistical average) becomes acute when Claude is shaping your thinking at the level of project design. This risk doesn't feel like a risk; it feels like productive collaboration.

To reach Level 3

Context-Aware Collaborator → Workflow Integrator

Conceptual Shift

"AI can act on files, not just respond to text."

Claude Code can read, write, and transform actual documents in your project — not copies you paste into a chat window, but the files themselves. This is qualitatively different from chat. And repeatable process design is different from conversation design: a workflow specifies what is the input, what is the expected output, what does a good result look like, where and how do you verify.

Experiment

Pick one file-heavy task that recurs in your work — summarising a set of papers, maintaining a structured bibliography, applying a consistent annotation schema to transcripts. Do it once with Claude Code. The goal is not efficiency — the first time will be slow. The goal is to understand what working at the file level feels like and where verification becomes necessary. Pay particular attention to the moment you would normally just accept the output: that moment is where your verification step belongs.

Practice

Build verification into every workflow before you rely on it — not as an afterthought, as a designed step.
Document your workflow decisions: what prompts you used, what you changed, what you decided was out of scope.
Watch for false systematicity: repeatable workflows create an impression of rigour that the underlying outputs may not deserve.

Note. Level 2 is a strong, defensible endpoint for most humanities researchers. Consider whether you have a specific use case that genuinely requires Level 3 before moving forward.

Next level →

Workflow Integrator

← Level 1: Regular Conversation Partner