A friend asked me how to actually build a company that runs on AI agents.
I drew him 4 simple diagrams and this is what I told him:
For this to work, a few things have to be true.
- The humans move up to strategy, taste, and judgment while agents handle the execution.
- The whole business becomes readable to agents. Your data, SOPs, pricing, permissions, and decisions all live in one shared context layer.
- And you point it at the right work. Repetitive enough for an agent, complex enough that the incumbents never bothered. That's the goldmine.
In the old world, the company was the people. They held the knowledge, made the calls, did the work.
In this new world, the people become the creatives, the agents become the labor, and the company itself becomes the context layer.
That shared brain is the actual company now. The humans and the agents are just plugging into it.
Which means the most valuable thing you can build in 2026 is a business so well-documented that an agent can run it.
I see it everyday with @MeetLCA. I don't talk about it much publicly, but we've built a SWAT team for building AI-native orgs and AI-native products.
The moat is how legible your company is.
I drew it all out below.
African startups should stop defaulting to AWS and Kubernetes.
Get a VPS on Hetzner, deploy with Coolify, and start experimenting until you start making money.
Oh, and when you do start making money, still stay there!
My year in poetry according to #ChatGPT:
You built systems from spark into flame,
Turning chaos to order by name.
From checkout to quote,
Every snippet you wrote
Showed the craft and the calm of your game.
I love it!
“Inside OpenAI, DeepMind, Meta FAIR, and Anthropic, teams that once focused exclusively on scaling LLMs are now pivoting toward protein folding, gene-expression modeling, molecular simulation, enzyme optimization, and lab automation.” — Audrey Xie https://t.co/dnf1V2a8jk
Apparently this video has all of X in a frenzy. If it had come out before the AI era, people would be fawning over it as great art, but now they are so clicker trained that any mention of AI sends them into a verbiage frenzy and they anoint anything AI related as slop.
Thought I knew everything about preloading assets in HTML.
Today I discovered this little bit: onload="this.onload=null;this.rel='stylesheet'"
Apparently, this helps with that pesky render blocking problem - who knew?
#buildinpublic
In the last week alone I've created a bunch of mini apps. All very useful. Here's the ones I like:
1. Github2Blog - turn your Github pushes into blog posts
2. Upwork Job Handler - score upwork jobs
3. Botdump - capture emails from bot spam
My favorite so far:
4. Code Visualizer - see the connections within your code
What did you conjure this week?
#buildinpublic
Something my brother and I have been building over the past few months: CapitalFlowAI - a cognitive AI trading engine that delivers unique market insights and will power our Brokerage service.
Like all my projects, my focus is on user experience, clean launches, smart automation, and scalable content delivery.
Behind every feature we build, there’s a system making sure it ships fast, scales easily, and supports dynamic scenarios.
Our latest system update includes improvements across newsletter generation, blog publishing, and deployment pipelines. That means more insights, delivered faster and with less human bottleneck.
The result: deep insightful development articles detailing our latest improvements.
👉 Read the latest on enhancing short position strategies and portfolio insights:
#trading #investing #stockmarket
Mastering Market Moves: Enhancing Short Strategies and Insights https://t.co/u5eh680woi
Wanted to convert a brand icon from PNG to SVG.
All the free tools I tried failed to provide a perfect version. Had no choice but to edit in BoxySVG
Glad I did, I'm picking up new skills. Thinking I can make custom vector icons now.
There's always value to be gained from practical learning.
#buildinpublic
I asked Github CoPilot to write a new README for an old app I had on the shelf.
Copilot understands my app better than I do 😃
... but, just a little better. #buildinpublic
Attribution is screwed, and it will get worse in this zero-click world filled with LLMs.
We had 400 businesses survey their customers on how many channels they interacted with the brand on before they purchased.
People’s memories aren’t always 100% accurate, but it directionally gives you a sense.
It doesn’t matter what the price point is or whether it is B2B or B2C, people interact with a brand on at least 3 channels before they purchase.
In most cases, they interact multiple times on that channel… so the touch points are even greater.
Don’t assume that it isn't producing results if you can’t track it. And if you can track it, don’t assume it is accurate.
A great example of this is that I just bought a Matic Robot Vacuum. My buddy Eric Siu first told me about it… then my sister showed it to me at her house last week and said I should get it, and then Eric showed me an X thread from the Shopify CEO on how Matic is great… I dug in and read the comments on that X thread…, and then I purchased it.
It cost me $1201.77, including shipping and taxes. I didn’t want to rush into purchasing something that expensive.
Now, how is Matic going to attribute that sale? I went directly and purchased…, and there’s no way for them to know what caused the sale.
On and off, something I've been working on for the @MagicRoomApp :
A ceiling-based sky environment ✈️😋
Still WIP... planned to add more elements like rainbow, birds, etc.
Should I finish it? 🤔