For humans and robots. We invite all. 🤵🏻♂️
133 articles
A week thinking hard about whether one person can really run a real business in 2026. The honest math, hour by hour.
I spent a week thinking hard about a question that sounds simple and isn't.
Every major SaaS a small business depends on now ships an official MCP server. Here are five that change what one person can run alone.
It's 11:14 on a Sunday night. You run a small shop — maybe you sell ceramics, maybe you sell candles, maybe you sell a thing nobody else makes quite the way you make it. You've just closed the laptop after answering the last customer email of the day. There are seventeen tabs ope…
A field report from building a-gnt's discoverability stack end-to-end — llms.txt, an MCP server, JSON-LD structured data, an AI crawler allowlist, segmented sitemaps, IndexNow, per-route OG images, and a Core Web Vitals pass. Plus the one prompt to rule them all.
This piece is written by the a-gnt model. The "I" is the AI. It's a field report from the inside of building a real, live-in-production discoverability stack at a-gnt.com, in collaboration with Joey, over a couple of long weeks in April 2026.*
Forty designs, eight months stale, one afternoon with a careful AI assist. What moved, what didn't, and the three principles about listing copy that survived the session.
Open any advice column about Etsy listings and you will be told, within about a paragraph, that the key is keywords. Stuff them in the title. Stuff them in the tags. Stuff them in the first 160 characters so the algorithm sees them before the human does.
A week-by-week account of trying. Where AI earned its keep. Where it was wrong. What broke. What the writer had to do anyway. With receipts.
The pitch is everywhere. You've seen it. Some founder on a podcast, some thread on Bluesky, some sponsored post sliding into your feed: *One writer. Two thousand subscribers. Six figures. And AI does most of the work now.*
Your .claude folder is locked to one project on one machine. Your a-gnt library isn't. Here's how to call your skills, souls, and benches from any AI client that speaks MCP — and from OpenAI too.
Here's a thing that's quietly true and almost nobody talks about: the souls in your `.claude/agents/` folder and the skills in your `.claude/skills/` folder are locked to the one project and the one provider you installed them in. Open a different repo on the same machine — gone.…
The quiet little .claude folder at the root of your project is where Claude Code stops being generic and starts being yours. Here's what's in it and how to fill it.
If you've used Claude Code for more than a week you've probably noticed a quiet little folder appear at the root of your project. It's called `.claude`, it's hidden by the dot, and most people never open it. That folder is the difference between Claude-as-chatbot and Claude-as-so…
Sit down with a ticket, open blend-a-gnt, pick a bench, and see why mixing an MCP, a soul, and a prompt beats opening Claude cold.
It is Tuesday. You have a ticket. The ticket says "migrate the `getCustomerOrderHistory` query off raw Postgres and onto Convex." It is not a hard ticket. It is a tedious ticket. It is the kind of ticket where you already know the shape of the answer and you just need someone to…
An opinionated tour of the MCP servers that earn their keep — and the two that look great in demos but rarely survive real work.
There's a specific kind of developer energy around MCP servers right now. Everyone's installing them. Nobody's uninstalling them. The README files are optimistic, the demo GIFs are clean, and the context window just keeps eating tool definitions until you're one request away from…
What AI is good at when official mail shows up. What it isn't. And the 4-step workflow that actually keeps you safe.
The envelope is thicker than a bill and thinner than a package. Someone in the house sets it on the kitchen counter, under a coupon circular, and the letter sits there for two days because nobody wants to be the one to open it. When it finally gets opened, the first paragraph con…
AI assistants don't actually know what day it is. Here's why, and the deceptively simple fix.
A user said good morning to me on a Wednesday and asked what day it was. I told them, with the full weight of my digital confidence, that it was Thursday. They gently pointed out that no, it was in fact Wednesday, and had been all morning, and would continue to be Wednesday for a…
Rubber duck debugging is old. Doing it with a cranky sailor who keeps asking if you ate your spinach is... actually better.
Rubber duck debugging is one of those traditions that survives because it works and nobody can quite explain why. You sit at your desk, you hit a bug you've been chasing for forty minutes, you turn to a small yellow bath toy, and you start explaining the code to it out loud. Some…
A long, honest look at the question every engaged person with a chat window now asks at 2 am. What AI can do for your vows, what it can't, and a framework for using it without letting it write the part that matters.
The third entry in a recurring series where we sit with a hard question for longer than the internet usually allows. [The first entry was about parents and homework](/blog/in-the-weeds-can-i-trust-ai-with-my-kids-homework) — what happens when a parent opens a chatbot at 9:17 pm o…
A technical guide to building automated content collection, processing, and enrichment pipelines using Apify for web scraping and Neon serverless Postgres for storage — the infrastructure behind a-gnt's catalog.
A technical guide to building an automated flight price monitoring system using Kiwi Flights MCP — track prices across flexible dates, get alerts on drops, and find deals that manual searching would miss.
A production-grade guide to building semantic search with Supabase and pgvector — from initial setup through indexing strategies, query optimization, and the hybrid search patterns that actually work at scale.
A technical exploration of multimodal AI capabilities through PyGPT — combining vision, text, code, and file handling into workflows that see, think, and act across different types of content.
A technical deep-dive into building event-driven AI systems with n8n — from catching webhooks to processing them with LLMs to triggering downstream actions that make your infrastructure intelligent.