I went quiet for three months.
No posts. No comments. No engagement. I let my LinkedIn sit. I stopped opening X. I closed the loops that had been training me to think in headlines and posts.
I did not announce it. I did not write a "going dark" post. I just stopped showing up.
For three months, the entire output of my day went into one thing. Building a startup with two co-founders.
The two people I disappeared with
Virgil, my co-founder, is one of the best marketers I have ever worked with. He thinks in audiences, hooks, sequences, offers. He can sit on a call with a founder for 30 minutes and surface the exact angle the founder has been circling but cannot name. I watched it happen 20 times.
Victor, my other co-founder, runs a 7-figure ad agency. He has been spending other people's money on Meta, Google, and TikTok for years. He reads a campaign dashboard the way a doctor reads an X-ray. The whole room is looking at numbers. He is looking at the leak.
Both of them have spent more time inside the messy, emotional, fast-moving day of an agency than I have spent inside any one thing in my life.
I went quiet so I could sit next to them while they worked. Not to give advice. Not to architect anything. Just to watch and ask questions until I understood the shape of the business they live in.
Three months. Heads down. No audience.
What I learned that I did not know existed
Some of this I could not have written three months ago. I had read about it. I had heard people on podcasts say it. But until you sit inside the work, you do not actually know it. You just have words.
How agencies actually work. The pitch deck version is "we run ads for clients." The real version is a constantly shifting balance between client expectations, platform changes, account access, creative production, reporting, retention, and the emotional state of one founder at a time. Every client is a different business. Every business is a different shape of trust.
How different every client really is. Two clients in the same niche with the same ad budget can need completely different strategies because their unit economics are different, their offer is different, their sales process is different, and the way they talk to their audience is different. There is no template. There is only learning the specific business in front of you, fast.
How emotional this business really is. Clients are not buying ads. They are buying belief in their own business at a moment when belief is shaky. Wins are celebrated. Losses are felt personally. The agency that survives is the one that handles both with the same steady hand.
How to read campaign data and know in 30 seconds what is working and what is bleeding money. I used to look at a campaign dashboard and see numbers. Now I see a story. Where the spend is going. Where the conversion is breaking. Which audience is carrying the result and which one is dragging it down. Where the leak is. That skill is not in any course. You learn it by sitting next to someone who has spent years doing it on real money.
How leads leak between the click and the sale. A click is not a customer. There are 6 places in between where you can lose them. The ad copy. The landing page. The form. The follow-up. The objection handling. The close. Most agencies optimize one of these and miss the other five. You only see all six when you live inside the funnel for months.
How ROAS, CPL, and conversion rates tell a story most people miss. These are not metrics. They are sentences in a longer paragraph. ROAS without CPL is meaningless. CPL without quality of lead is misleading. Conversion rate without time-to-close is a trap. The story is always in the relationships between the numbers, not the numbers themselves.
How to make your content visible to ChatGPT, Claude, and Perplexity, not just Google. Search did not stop at Google. It moved into AI chat. The way your site is indexed, the schema you publish, the citations you build, the way an AI model decides to mention your name in an answer. This is a new layer. Most businesses are not playing in it yet. The ones who start now will have a real lead.
How skills and agents are making 3-person teams operate like 30. Claude Code with skills. Hermes-Agent running background work. Custom agents that handle the boring parts of an operation. None of us are working 100-hour weeks. We just removed the friction from the parts where 30 people used to be needed.
How OpenClaw and Hermes are shaping the future of AI coding. This one is the most exciting to me as a coder. The agent layer is moving fast. The tools that win the next 12 months are the ones that compose well, run reliably, and let a small team ship like a big one.
What three months of silence actually felt like
It felt boring in the first week.
I kept reaching for my phone to check what I had missed. I had not missed anything. The first week is the hardest because your brain is still running the old reward loop. Open the app. Get the dopamine. Repeat.
By week three the reach was gone. I stopped wondering what was happening on X. I stopped imagining a post that was not getting written.
By week six the silence had texture. I was thinking deeper about problems. I was sitting longer with things before reaching for a tool or a tweet. The internal monologue got quieter, then sharper.
By week ten I was not the same builder I had been when I started.
I learned things I did not know existed. From building next to people who do it every day.
That is the trade. You give up the feed. You get back depth.
What I would tell anyone thinking about doing this
If you have been thinking about going deep on something, this is the part nobody tells you.
You do not need a sabbatical. You do not need to quit your job. You do not need to announce anything. Three months and one room is enough.
Turn off the feed.
Stop consuming.
Start building.
The version of you that comes back will not recognize the version that left.
What would you build if nobody were watching?
