The first thing I tackled was getting three properties onto identical CRM systems. AI Officer, CAIO Coach, and a third site. The idea was simple. Build the CRM plan once, copy the markdown file to the others, and have a consistent structure across all three.
It mostly worked. CAIO Coach came online smoothly. But AI Officer has two properties and when I tried to consolidate them into one database, the AI Officer side broke. CAIO Coach works fine. AI Officer does not. I still do not know why.
That is the reality of building fast. Sometimes things just do not cooperate, and you do not have time to debug because you are walking into a conference in 90 minutes.
In roughly an hour, I set up a CRM system (mostly), team messaging via Telegram, and automated email notifications. One person. No developer. Telegram is the best platform if you want to automate messaging. Today I proved it. These are the communication rails that let a one-person company feel like a team.
The Content Machine Takes Shape
Jan is leading the content production work, and tomorrow is the big push. We are going to completely finish our marketing team and get it live with fresh content starting Monday.
Here is the plan we locked in. We chose five blog styles to rotate through. Things like listicles, how-tos, and a few others from our catalogue. Then we defined two core data sources for topics.
First, the book. Bill Hajdu's content. The Fire Pig, the astrologer. A deep well of material. Second, weekly interviews with Bill. One conversation per week, recorded and converted into content.
From those two sources, we will produce two main blog posts per week. Then five days of social media content gets generated from that same data. Two inputs, seven outputs. That is the ratio we are building toward.
Getting the Data Right
I went deep into Harvard's persona frameworks today and landed on a blend of two approaches: the Empathy Map and Jobs to Be Done.
The AI actually recommended the mix, and it makes sense. Jobs to Be Done tells you what your customer is trying to accomplish. The Empathy Map tells you what they are feeling while they try. Together, you get a customer profile that is specific enough for a writer, human or AI, to actually write to someone, not just about something.
This is the piece most people skip. They jump straight to writing without building the data layer underneath. But good data means the writing is almost perfect. The richer your customer profile, the less editing you do on the other side.
We now have a targeted customer profile feeding into the writer agent, sitting on top of two defined content sources, outputting across five blog styles. The data sets keep getting richer, and that is the whole game.
The Pipeline Grows
I picked up three people at my event today who said they want to come to the retreat. That puts us at eight confirmed. Eight times three thousand is $24,000 in the pipeline.
Not bad for a one-man company on Day 6.
The conference was a full day. But the morning infrastructure work meant the systems kept running while I was shaking hands. Telegram notifications. Email confirmations. CRM capturing leads. The leverage compounds.
What We Actually Shipped
The compact version. No hero lap, no drama, just the list.
- CRM systems across three properties. AI Officer, CAIO Coach, and a third site. CAIO Coach works. AI Officer has a consolidation bug. Tomorrow's problem.
- Telegram team messaging. Set up and running. Absurdly easy to automate.
- Email notifications via Resend. Automated and live.
- Content strategy locked in. Five blog styles, two data sources, two posts per week, five days of social content. Two inputs, seven outputs.
- Harvard persona frameworks applied. Empathy Map + Jobs to Be Done feeding the writer agent.
- Writer agent nearly complete. Persona data, content sources, and blog styles all connected.
- Three new retreat confirmations. Eight total. $24K in the pipeline.
- Nano Bananas looking good.
What's Next
Full work day tomorrow. Finish the marketing team. Get fresh content queued for Monday. Debug that AI Officer CRM issue. The sprint continues.
The Real Insight
The lesson is about data layers. Everyone wants to talk about AI writing content. Nobody wants to talk about the boring part. Building the persona, cataloging the source material, choosing the frameworks, defining the styles.
But that boring part is where the leverage actually lives. An AI writer with bad data produces generic content. An AI writer with a Harvard-backed persona framework, a defined content source, and five proven blog styles produces content that sounds like it was written by someone who knows exactly who they are talking to.
Because it does.
How did you set up CRM, Telegram, and email notifications in one morning?+
What persona frameworks work best for AI-generated content?+
What is the content ratio you are building toward?+
How did you get $24K in the pipeline at a conference?+
Dave Hajdu is the founder of the AI Officer Institute and Edge8 AI. He works with founders and executives across more than 20 countries to build the leadership capabilities the AI era demands. Learn how to build your own AI team at caiocoach.com.