The 8-week program

Eight weeks of agentic AI fluency written for pharma operations. Each week asks one question a leader has to answer, in short units you work through at your own pace. The first two units are open to everyone.

Join the Collective

Units marked Members open when you join. It takes a minute to apply.

Week 1 · Literacy

What is this, really, and where is the money?

When someone asks you to fund AI, can you tell what the system will actually do, and which of your costs it is meant to move?

Week 2 · Knowledge

Can we trust what it tells us about our own documents?

When a team proposes "chat with our documents", can you tell whether its answers will come from the effective version, show where they came from, and reveal only what the person asking is allowed to see?

Context is the job Members
How retrieval works, and how it fails Members
Pick the tool for the job Members
Where knowledge work hurts Members
Make it cite Members
Questions for Monday Members
Page 2: knowledge targets Members

Week 3 · Action

Which processes should agents touch, and where must a person decide?

When a team proposes an agent, can you tell whether the process needs one, what it may do on its own, and where a person must decide?

Workflow or agent? Members
Tools, MCP and what can go wrong Members
How long can it work alone? Members
The deviation investigation, redesigned Members
Be the approver Members
Questions for Monday Members
Page 3: the process shortlist Members

Week 4 · Context

Is our data ready for agents, and who owns its meaning?

When an agent joins records from your MES, LIMS, ERP and quality system, will it know they mean the same thing, and who in your organisation decides what they mean?

Same word, three meanings Members
The vocabulary ladder Members
How the plant's systems connect Members
The readiness interview Members
Ask it to join two sources Members
Questions for Monday Members
Page 4: data-readiness questions Members

Week 5 · Trust

What may the AI decide, who signs, and what will regulators expect?

Before you fund an AI use in a regulated operation, can you say what it may do alone, who signs for what it produces, and which rules actually apply today?

The line that isn't there Members
The landscape, dated Members
The human in the loop is weaker than you think Members
The vendor updated the model Members
Red-team your own assistant Members
Questions for Monday Members
Page 5: decision rights and risk appetite Members

Week 6 · Scale

Which bets, in what order, and what happens to our people?

You can't fund everything. Which AI bets go first, how will you know within 90 days whether each one works, and what happens to the people whose work it changes?

Read the evidence like a CFO Members
Rank the portfolio Members
Who gains, who's stranded Members
The pilot that won't die Members
Write the hypothesis, then attack it Members
Questions for Monday Members
Page 6: portfolio and value case Members

Week 7 · Run

What will we be asked to fund, and will it still work in six months?

When the pilot is over, what will it cost to run every month, how hard would it be to leave the platform, and who keeps it working?

The cost of a successful task Members
Lock-in moves, it doesn't vanish Members
It worked in the demo Members
The platform pitch Members
Ask it what it costs Members
Questions for Monday Members
Page 7: adoption and organisation plan Members

Week 8 · Commit

Would you fund it? Make the seven-minute case.

Seven minutes with the executive committee and a red team from quality and finance: is your AI case one you would fund yourself?