Notes from operating agentic companies.
Agentic companies, lean operation, process-to-agent, founder craft. Written answer-first, by the team that does the work.
The art of the commercial approach
A fast first reply, a visit with a checklist, a complete proposal, a short presentation and two follow-ups are not good manners. They are how a customer who cannot judge your work decides what it is worth, and the margin follows that judgment. The research on how people read price, design and trust, and what winning proposals actually contain.

Notes from operating agentic companies.

Your week has forty hours of work in it and sixty hours of business
Answering the phone, keeping the feed alive, finding a slot for an estimate, writing the quote, chasing the invoice. None of it is the job the customer pays for, and it is where most of a small-business owner's week goes. What the surveys actually measured, where the government data disagrees with the vendors, and which bucket to hand over first.

The customer called while you were on the roof. Nobody answered.
Every service owner I interview tells me the same three stories: the reply that came too late, the quote that went out too late, the day lost confirming what was already agreed. Here is what the sources actually say those three hours cost, which famous numbers I refused to use, and why a company with seventy salespeople bleeds in exactly the same place.

Agents don't hallucinate on their own, they get overloaded
Agents rarely hallucinate because the model is broken. They hallucinate because one chat was asked to be memory, filing cabinet and five specialists at once.

AI companies are starting to behave like unions of digital workers
AI companies now organize, tier and price a new kind of labor. Companies run on headcount plus token count and agent count, and leaders, managers and professionals all have to relearn the job.

The zero moment of truth moved inside the answer
Google named the moment your customer decides, back when deciding meant searching. That moment moved into the AI answer, and most small businesses are not in it. Here is what actually changed and what a small business can do this month.

Your agent will only be as honest as your records
Six years ago I wrote that having customer data is not the same as having it organized. The argument has not changed. The stakes have: disorganized records used to cost you a follow-up, and now they make your agent confidently wrong in front of a customer.

Give your first agent the boring job, not the painful one
I use four filters to choose the first process an agent takes over, the same filters I have used for years to find the first thing worth systematizing. The process that hurts most is rarely the one that is ready.

AI can write your words. It cannot decide what you are worth.
Answer fast: what is your value proposition? I have asked owners that question for over a decade, and the answer is usually a price or a list of features. AI made the words free to produce, which means fluent copy stopped being proof that anyone did the thinking.

An agent belongs at the top of your funnel, not the bottom
How many new contacts does your business make in a day? Six years ago I argued that a funnel is a flight plan, not a formality. It is also a process, which means the four filters I use to pick a first agent can be run against it stage by stage. Three of them stop at the same place.

You do not have a lead problem. You have a first-reply problem.
Most small businesses already get more inquiries than they close. What they are missing is not traffic, it is the first reply. I wrote this two years ago and the argument still holds, but the answer changed: the first reply is now the one piece of selling a business can genuinely hand to an agent.

The mistake is not being absent. It is presence that never reaches selling.
Being on Instagram was never the problem, and being absent was never the expensive mistake. The expensive one is a presence with no wired connection to how you actually sell. I spent eight years building that connection inside an industrial company, and AI has just made the gap cheaper to widen.

Two years of AI use cases, and the three traits the survivors share
I wrote a list of generative AI use cases by industry two years ago. Most of what was on it is now either ordinary or forgotten, and the split between those two outcomes was not about the industry at all. Here is the updated list, and the pattern that actually decided which ones lasted.

