I completed Anthropic's Claude 101 and their AI Fluency: Framework & Foundations because most 'AI for work' content is prompt tricks. Claude 101 gives me the mental model of the tool itself - prompts, context, output, workflows - and AI Fluency treats collaboration with a language model as a professional practice with four disciplines: Delegation, Description, Discernment, and Diligence.
What I hand to Claude
- First drafts of outbound sequences, follow-up emails, and status updates
- Rewriting long threads into short, next-step-first summaries
- Cleaning CRM notes into a consistent structure
- Turning unstructured client feedback into a tagged list
- Explaining a tool's edge case to me before I ask a client about it
What I never hand to Claude
- Anything a client hasn't agreed can touch an AI system
- Numbers I haven't independently checked - reconciled cash, forecast, headcount
- Any client-facing artefact without a human review step before it ships
- Confidential data that would breach a privacy or compliance boundary
The verification step is the whole job
AI Fluency's Discernment discipline is the one that matters most in operational work. It is not enough for a draft to sound right - the operator has to be able to point to the source for every non-trivial claim, name, number, or date. If I can't, the draft doesn't ship.
Why this is a competitive advantage, not a liability
The teams I work with move faster because Claude is in the loop, and they trust the output because the loop always ends with a human. That combination is rarer than it should be - most 'AI-assisted' operational work is either slow or unverifiable. It doesn't have to be either.
If you want the credential context: both certificates live on this site - Claude 101 and AI Fluency: Framework & Foundations - with the syllabus, the certificate PDFs, and the workflows behind each one.