Connections in Math: the two kinds of random
Comments
Insights
Practical writing on software architecture, SaaS products, AI automation, legacy modernisation, and the business of building reliable systems.
Curated links from external sources — not 360Softy original articles.
Comments
I recently left my role at Superlogic. I am not leaving technology. I am not stepping away from Instead, I am using this next chapter to focus on my family, That last part is the experiment I am most interested in. My kids are growing up in a world where the cost of turning But faster development does not automatically produce better AI can generate code, designs, copy, research, and business Those are the skills I want to teach. ## The goal is not to teach them prompting I am not trying to trai
If you've heard about the Model Context Protocol (MCP) but aren't sure how to build something with it, this guide is for you. In my previous article, I explained what MCP is and why it matters. If you haven't read it yet, I recommend starting there first. Understanding the concepts is one thing. Building an AI agent that actually uses MCP is another. In this guide, you'll build a simple AI agent that communicates with an MCP server, uses external tools, and returns useful responses. More importa
If your app needs to send a welcome email, a password reset link, or an order confirmation, you don't need to stand up a mail server. You need one HTTP call. This tutorial builds a small Python CLI that sends transactional email through Mailgun's API, then checks whether Mailgun actually delivered it. By the end, you'll have replaced "sending email" with a simpler model: call an API, read the response, done. Before you start, gather the following: Basic Python knowledge. Python installed. A Mail
Here's the thing: i built LLM pipelines for a mid-stage fintech before joining Global API's solutions team. Every quarter the same argument came up: do we go direct to OpenAI, route everything through Azure, or layer in an aggregator? What I learned over those months is that the "right" choice depends almost entirely on your failure tolerance — and most teams dramatically underestimate theirs. This post is the matrix I wish someone had handed me on day one. The piece of conventional wisdom that
Every watch-price tool I found gives you the same thing: a pile of raw listings. Scrape Chrono24, get 60 JSON objects, and now you write the median logic, you filter out the "strap only" listings poisoning your stats, you figure out that the first cards on the page are promoted ads for a different model entirely. I wanted the opposite: one call in, one market analysis out. So I built it. Ask for "Rolex Submariner 16610" and you get back a single structured answer: { "analysis": { "n": 46,
Article URL: https://arxiv.org/abs/2605.20049 Comments URL: https://news.ycombinator.com/item?id=48798815 Points: 190 # Comments: 89
The problem: you don't own a store, you rent a stack If you have ever run a shop on a hosted e-commerce platform, you know the feeling. You sign up for a tidy monthly plan, and within a fortnight your "cheap" store is quietly bleeding money. Reviews? That's an app. Loyalty points? Another app. A referral programme, a proper blog, multi-currency, subscriptions, a help desk, B2B quotes? Apps, apps, apps — each one £15–£40 a month, each one another dashboard, another dependency, another thing tha
Work with 360Softy
Book a free consultation and we will tell you honestly whether we can help.