How We Design Workflows
We ground every workflow in validated business and academic frameworks, not prompt tricks. From that foundation, one method carries each page in this section: plan the program with the 5Ds, document the workflow in seven elements, assign each element to a human or a machine, test the document, then ship through three stage gates. It is the same method we teach in the AI Officer certification, applied to our own company.
The invisibility problem
Every organization runs on workflows nobody has written down. They live in the heads of the people who run them, which means they stall when that person is out, they cannot be improved because they cannot be seen, and they can never be handed to AI. The work your team relies on most is usually the work that is least visible. Making it visible is the first job, and it is a leadership job, not a technical one.
The 5D Program Brief
No workflow gets built without a one-page brief covering five Ds. It keeps the program honest: a real problem, real data, a documented flow, a determined return, and a deployment plan.
Write the problem statement and a FAST goal. A workflow that cannot name its problem is a solution looking for one.
Name every piece of data the workflow needs, where it lives, whether it is clean, and whether AI can reach it.
Document the workflow end to end and draw it. The pages in this section are this D, published.
Put a number on what the workflow saves or earns. If the return cannot be determined, the program does not start.
Plan the rollout through the three stage gates: prototype, pilot, production.
Seven elements, every time
A workflow is not a paragraph, it is a structure. We break every one into the same seven elements. The assignment chips below show the typical split; every workflow page in this section carries its own map.
The event that starts the workflow. A form submission, a calendar date, a message. Named precisely, or the workflow starts on vibes.
Everything the workflow consumes: documents, records, context. Each input names where it lives and who fetches it.
The judgment calls inside the flow, written as rules explicit enough that a new hire, or an AI, applies them the same way twice.
Where work goes after each decision, including the exception paths that usually live in one person’s head.
The artifact the workflow produces: a document, a record, a payment request, a ranked list.
How the output reaches the people who need it: a notification, a dashboard, an email, a published page.
What the workflow tracks about itself: cycle time, hit rate, follow-through. Untracked workflows cannot improve.
The Centaur Map and the New Hire Test
Centaur Map: human, machine, or both
- Every element gets an explicit assignment based on comparative advantage
- Machines take triggers, routing, delivery, and measurement
- Humans keep judgment with consequences: approvals, hires, money
- Where the machine needs data it does not have yet, the map names it
New Hire Test: the quality gate
- Someone who has never run the workflow runs it from the document alone
- Every place they stall is a gap in the document, not in the person
- The weakest step is always an undocumented decision rule
- If a human cannot follow the document, neither can AI
Three stage gates
A documented workflow earns its way into production. It does not get declared into it.
The frameworks behind the workflows
The workflows in this section do not float free. They draw on a library of business frameworks we teach in our leadership and AI Officer programs, each one producing a dataset that AI can work from.
Feeds the 1-1 coaching workflow with how each leader actually works.
Personal datasets that let AI coach and draft in your own voice.
The frame behind hiring, onboarding, and the resume screen workflow.
The structure inside every 1-1 conversation and coaching profile.
Goals connected from company to team to individual, reviewed on a rhythm.
Stay interviews and early-warning signals, run before notice is handed in.
One real decision, framed properly, before any dashboard gets built.
The seven-element document behind every page in this section.
Problem to tested prototype in one sitting, ending in kill, iterate, or scale.
How a new workflow actually lands with the people who must adopt it.
Prompt frameworks are dead
Learn to apply real, tested academic and business frameworks to the problems you are trying to solve.
This method is teachable. And buildable.
We certify leaders in this method through the AI Officer program, and we build these workflows directly for clients. Either way, your invisible processes become systems.