The Fractional COO: From Drafting to Doing
I am looking at a blank Google Doc, but I am not staring at it. I am architecting it. Beside me, on my second monitor, is a Gemini window. We are not chatting about the weather or asking for a recipe. We are in the middle of a clinical workflow automation.
"Gemini — I've just finished the clinical intake script for the new PWA. I need you to write the Apps Script that will take the JSON output from the app, map it to the Patient Records BigQuery table, and if the Risk Level is marked as High, trigger an encrypted Priority notification to the clinical lead's Google Chat."
The most consequential shift in the relationship between professionals and AI is the move from conversation to cooperation.
For years, our relationship with AI was conversational. We asked, it answered. It was a drafting partner. In 2026, we have moved into the age of agentic AI — and for the solopreneur and the small NFP leader, this changes the operating economics of the business. AI is no longer just writing the email. It is doing the work. It is the middle-management layer you previously thought you could not afford.
The "Drafting" Trap
Most people use AI as a super-powered typewriter. They get it to draft a social media post, a grant response, a clinical summary. Then they copy. They paste. They edit. This is still janitor work.
While you are copying and pasting, you are still the human in the loop for a low-value task. You are paying the Complexity Tax with your time. If you have to touch a piece of data twice, you have a broken workflow.
The Fractional COO approach means the AI does not just write the clinical summary. It processes it. It identifies the action items. It updates the project tracker. It schedules the follow-up meeting. It files the document in the Sovereign Brain (Essay 11).
The Agentic Architecture: Apps Script as the First Employee
To move from drafting to doing, you need a bridge. In the Microsoft world, that bridge is Power Automate — powerful, but heavy, and often locked behind premium licensing. In the Google world, the bridge is Google Apps Script.
Apps Script used to be the hidden basement of Workspace. It required you to look at a code editor. Gemini has democratised the engine room.
When I built the automated voiceover flow for my healthcare courses, I did not hire a developer. I used Gemini as my fractional CTO:
- I described the logic in plain English.
- Gemini wrote the code.
- I implemented it inside my Workspace tenant.
Four hours of work that saved me forty hours of manual processing per course. I was not drafting. I was architecting a machine.
The economics here echo the no-code argument from Essay 10. Kok et al. (2024) and the broader research on low-code AI for SMEs document the same pattern across multiple platforms: SMEs and solopreneurs who adopt agentic automation see meaningful operational throughput gains without increasing headcount, because they have eliminated the translation error between business reality and technical implementation.
The "Sovereign Workbench": n8n and the Multi-Cloud Reality
Even when you go Google-native, you still live in a multi-cloud world. Your patients may use a different booking system. Your donor data may live in a specialised CRM. To be a true Fractional COO, you need to cross the border between systems without paying a Complexity Tax.
This is where n8n earns its place. n8n is the fair-code, self-hostable workflow tool that fills the same niche as Zapier or Power Automate, but with sovereignty intact. Because it is self-hosted (or hosted in your own cloud), you own the data flows.
The pattern is familiar. A patient fills out a form. n8n sees the entry, asks Gemini to categorise it, then triggers actions in three different apps simultaneously — your Workspace, your CRM, your messaging tool of choice. The AI is not just helping you write. It is running your office.
Building Your Own Agentic Flows
The 2026 starting blueprint for the Fractional COO approach is small enough to build inside a single working week:
The intake agent. Create a Google Form for volunteer inquiries. Use Gemini to build a flow that monitors entries, checks your central calendar for orientation dates, and automatically sends invitation links.
The high-risk monitor. Use Apps Script to scan clinical intake forms for red-flag keywords. If found, have Gemini draft an immediate briefing for the clinician and send an alert via Google Chat.
Dynamic data ingestion. Use BigQuery's structured/unstructured connectors to process unstructured clinical data — handwritten notes, clinical images — alongside your structured spreadsheets, using Python or SQL directly within your Workspace environment.
Each one of these flows is a discrete agent. Together, they constitute a fractional operations layer that runs without you.
The Control Fear
For the solo operator, control is the security blanket. The fear is that letting go of the manual touch will lead to mistakes, hallucinations, or undetected errors.
The honest counter is that manual work is more prone to error than automated work. Your human error on hour eight of a mail-merge is materially higher risk than a well-tested agentic flow. The condition for letting go is governance — the secure foundation laid down in Essay 7 (BAA, Trust Rules, Sovereign Brain). You can let go of janitor work because the architect has built a secure foundation.
Reclaiming the A-Game
The gold light over the Macedon Ranges in the late afternoon used to be a signal that I needed to start the janitor work. Now it is a signal that my Fractional COO has finished the day's processing. The emails are sent. The tracker is updated. The Sovereign Brain is populated.
The self-taught enterprise is not about working harder. It is about working higher — moving the A-game out of the basement of manual tasks and into the strategic work that only you can do.
If you are still copying and pasting, you are still a janitor. It is time to step into the role of the no-code architect — using Gemini to build your bridges, n8n to cross your borders, and the cognitive surplus to do the work that only you can do.
