The Manual → Chapter 07

Operating in Public

I run my own firm as an operating system: nine AI subsystems doing the back office every day. This chapter is the working log of building it. The mistakes are left in, because that is where the lessons are.

Read in this orderStart with the first 3 below. They set up the rest.
08:0021:30
30 essays
07-Aug-2026
01Why I Run My Business as an Operating SystemI don't have a team. I have a system, and three businesses run on it. For years I thought growth meant more people, until I saw that most roles I'd hire for were a decision, a rule and a handoff. So I stopped hiring people to carry those rules and wrote the rules down instead.
13-Jul-2026
02Automated Tasks vs an Operating SystemTools execute steps, while an operating system owns the loops around them. This piece explains why counting automations tells a founder nothing, and what an installed operating spine changes.
20-Aug-2026
03What Founders Ask Before They Trust AI ToolsFounders have stopped asking what my AI system can do. They ask what it refuses to do without them, and whether it says so when it's unsure. Almost every tool can do a lot now, so what founders want to buy is a system whose behaviour they can predict.
13-Nov-2026
I Nearly Turned Away the Right ClientA buyer definition and a pain map are different tools. The segment where a problem screams loudest tells you where to aim, not who to accept.
06-Nov-2026
Which first job to give AI in a small firmThe best first job for AI in a small firm is a small, boring task the team repeats every week, where a person can check the answer in a few minutes and a mistake costs nothing. Sorting customer reviews, tidying a lead list and summarising the weekly report all fit.
05-Nov-2026
What your team should never paste into AIA small firm's staff should never paste five kinds of things into an AI chat tool: customer personal details, salary and bank records, anything a client shared in confidence, passwords and access codes, and notes about a named employee.
05-Nov-2026
Good Execution Kept Me from ChoosingExecution rewards speed and reliability, so you keep doing instead of deciding. The operator to builder transition stalls right here, because doing always feels like progress while a real choice feels like exposure.
04-Nov-2026
How a small firm can start with AI safelyThree steps for an owner who wants to try AI without a data leak: pick one boring job, write what the AI must never do, and have a person check its work.
29-Oct-2026
The Question Nobody Asked Me at Year FiveSkill at execution takes away the pressure to pick a direction, because someone who can finish anything never has to decide what is worth finishing. For me, the operator to builder transition began when I started asking what I would refuse to run.
22-Oct-2026
Devotion Is a ScheduleBecause motivation is a mood and moods move, while a fixed time on your calendar does not.
16-Oct-2026
Where the Pain Is Worst vs Who You Take OnA buyer definition and a pain map do different jobs. The segment where a problem hurts most tells a consultant where to aim, and says nothing about whom to accept. Anyone asking who needs an operating system consultant should answer by the problem in front of them, never by the size band where that problem is most common.
08-Oct-2026
When Software Says It Checked Its Own WorkYou don't know your automation did what it says until you check the checker. A report a system prints about its own work stays a claim until something that gains nothing from a pass has verified it. That's the quiet cost of moving from operator to builder: the founder stops inspecting the work and starts inspecting the report.
02-Oct-2026
Operator to Builder: My Own Report LiedYes, and it happens most to people who build their own systems. A tracker only knows what it was told, not what changed elsewhere. Somewhere in the operator to builder transition you learn to check the work against reality before you trust the record that describes it.
16-Sep-2026
Every Business Is the Same Seven BoxesYears of running operations in large businesses left me with one map instead of a pile of industry knowledge. Every business runs on the same seven parts, and the skill is knowing which part to distrust first.
21-Aug-2026
Why Every AI Tool I Build Has a Do-Not ListThe first version of every AI worker I built could act. The version I trust can refuse. Writing the do-not list took me longer than writing the abilities, because a worker that can't refuse is simply running without supervision.
14-Aug-2026
Why I Keep a Daily Log of What Went WrongI don't keep a file of what my business does well. I keep a file of every place it rubbed, and that's the file I open first each morning. Wins tell me what to protect, and friction tells me what to build next. #CompanyOfOne #OperatingModel #OperatorToBuilder
12-Aug-2026
Why I Cut Half of the AI Workers I BuiltI spent eight months fixing every problem by adding another AI agent. Then I deleted half of them in one weekend, and the system got better. Adding workers had been hiding a design problem, and removing them made me fix it.
05-Aug-2026
A Clean Report Built From No Data at AllLast week my automated system built a full report out of no data and reported success. One wrong character in a folder name led to an empty fetch, a confident wrong file and zero errors. Automation keeps the judgment in the work; it moves it to the moments nobody designed.
31-Jul-2026
Two Jobs Shared One Bot and Neither KnewOne of my systems went silent for two days without anything breaking. Two jobs were reading from the same Telegram bot, which lets only one reader in at a time. The new job took the stream, and the old one kept looking healthy while it heard nothing. The risk sat in the part they shared, which nobody owned.
29-Jul-2026
My Best Automation Does Nothing Most WeeksMy best automation does nothing most weeks. It reads one number, finds it healthy and shuts itself down without sending me anything. Building something that stays silent on purpose was harder than building the part that runs, because I had to define exactly what healthy means.
24-Jul-2026
Why I Let the AI Make Mistakes on PurposeI built my AI agents so they're allowed to make mistakes, and I did it on purpose. A mistake I can reverse, sitting behind a person's approval, costs me almost nothing. A perfect system I never ship costs me something every day, because shipping speed compounds and polish doesn't.
22-Jul-2026
Why I Keep the AI's Rules to Ten PagesMy AI operating system runs nine subsystems, and its rulebook is capped at ten pages. Every new rule has to beat an old one to get in. Adding rules never gave me more control; ranking them did.
17-Jul-2026
The Fix That Created a New FailureI wrote a rule so my agents would stop skipping a step, and they stopped skipping it. They started waiting instead, so the fix worked and the system got slower. The skipping was a loud failure I could see. The waiting was a quiet one, and I think the quiet one is worse.
13-Jul-2026
Can Your AI Show What It Did and Undo It?The trust problem in agentic AI is about provenance more than capability: can the system show where each action came from, and can the business undo it? This piece explains why decision trails and rollback proof decide whether a business can run on agents.
13-Jul-2026
Judge Consulting by the Founder Hours FreedDeliverables are the wrong measure of operating work. The honest measure for a founder is the number of operating hours freed each week, and this piece shows how to count them from two months of calendar.
10-Jul-2026
Why I Don't Fix a Problem the First TimeI don't fix a problem the first time it happens, because one occurrence can't tell me whether I'm looking at a pattern or a bad day. I log it the first time, investigate the second time and build the fix only on the third.
08-Jul-2026
How I Let AI Workers Earn Their AutonomyNo agent in my AI setup goes live on day one. Each one earns its freedom through clean dry runs, full visibility into what it did, and a watchdog that is already watching before it acts. The model is rarely the weak link; the missing gate usually is.
12-Jun-2026
Why Backups Alone Won't Keep You OperatingMost solo operators back up their files and stop there. What usually ends a one-person operation is losing the way back into an account, for example when one phone holds both the login and the recovery number. I audited my own stack for that and found two gaps.
07-Jun-2026
The 307 vs 308 Redirect Mistake I FixedA 307 and a 308 redirect land a visitor on exactly the same page, yet every machine reading the site takes them to mean opposite things. This piece covers the layer most site owners never test, and why ignoring it is getting more expensive.
03-Jun-2026
What My MBA Missed About Becoming a BuilderThe MBA taught me to name problems. A Rs.400 Cr factory taught me to feel them before they had a name. The degree made me articulate, and the years on the floor are what made me right.
If you want your team using AI safely

Start with one hands-on workshop for your team.

A hands-on session where your people build their first AI helper on their own work, with simple rules for using it safely.

Pick a time that suits you.

Book a call