Topic
8209 articles found
By default, an LLM forgets you the moment a conversation ends. Start a new chat and it has no idea who you are, what you told it last week, or what you prefer. For a real product, that is a dealbreaker. Users expect the app to remember. The standard fix is a memory pipeline you build yourself. Extra
You have an idea for a SaaS. You open Cursor or Claude Code, and the familiar ritual begins. Auth. A database. Payments. Transactional email. Permissions. The agent generates it. You spend the weekend reviewing and patching, you burn $100 to $300 in tokens on the same foundational plumbing every ser
My goal is to become a software architect. And the first step towards that? Becoming a really solid backend engineer. So naturally, in my infinite wisdom, I asked ChatGPT for project ideas. I know, I know, but hear me out, because this one's actually a gem. Build a fake bank. Banking systems force y
If you work in IT, Legal, or Procurement, you know the pain of the Vendor Security Review. A vendor emails you a 50-page PDF outlining their privacy terms, and you have to manually cross-reference their document against your company’s internal Standard Operating Procedures (SOPs). It is tedious, tim
Bilingual post · Post bilíngue Jump to: English · Português The best proof of a Pascal compiler is not another hello world — it is a small but real HTTP service. CrabPascal ships examples/crud/, a product CRUD API built with Horse, System.JSON, and generics collections. This walkthrough shows how th
A lawyer friend messaged me at 11 PM last week. He'd sent a 47-page contract to a client. Client signed it, sent it back. Everything looked good until my friend noticed page 12 had a typo in the payment terms — the kind of typo that, if a court ever looked at it, could mean the client owed nothing.
Open almost any linter config and you will find a setting like max-line-length: 80. Nobody in the room chose 80 on purpose — it was inherited, and the chain of inheritance runs back nearly a century to a piece of cardboard. The 80 in "80 columns" is the width of the IBM punched card. The card layout
Before working for 2 years on the Apache APISIX API gateway, I was mainly oblivious to API gateways. It's only by working with them that I understood their value. Decoupling the client and the server unlocks a lot of options: moving authentication to the API Gateway, securing APIs, deduplicating API
If you've built an AI agent that touches real enterprise data, you've probably hit this wall. Your agent pulls 2,000 records from Salesforce. Now what? The model can't reliably filter, sort, or group 2,000 rows inside its context window. You don't want to dump all of it as raw JSON. And spinning up
Both git merge and git rebase exist to do the same job — integrate changes from one branch into another. They produce different histories because they go about it in fundamentally different ways, and understanding that difference is what stops you from corrupting a shared branch. A merge takes the w
A few weeks ago I shipped NetDiag+, an iOS network toolkit — ping, traceroute, no private entitlements. I wrote up the ping/traceroute internals here. Then I spent a month adding 9 more tools. I expected "more of the same" — wrap a mtr keeps probing every hop and accumulates per-hop loss + RTT stats
GitHub officially switched to a usage-based billing model a few days ago. The era of unlimited $10 fixed fees is over 🥲 Recently, GitHub Copilot started introducing more detailed usage-based concepts around premium requests, model access, and AI consumption. Instead of treating all AI interactions
This is the next entry in a build-in-public series where I extract a real, production Laravel CRM into a reusable SaaS core, one module at a time, and ship each piece as a Composer package. So far: the foundation layer, auth on top of Fortify, multi-tenancy, roles and permissions, and the platform a
TL;DR: Amazon's Buy Box data requires JavaScript rendering—standard HTTP requests won't work. DIY Playwright scrapers fail at scale (35–55% success rate in 2026 conditions). Pangolinfo Scrape API is the production path: >95% success rate, structured JSON output, 5–15 min freshness. Complete working
Choosing the right domain extension in 2026 isn't just about availability. Your decision between .dev and .com shapes how developers, investors, and users perceive your brand. It affects everything from SEO performance to investor trust. The domain landscape has evolved dramatically. While .com stil
llm-cli-gateway 2.0.0 went out on 4 June 2026. npm now reports 2.0.0 as the latest version, and the public GitHub release carries the platform binaries, bundled installers, SHA256 checksums, release manifest, and Sigstore bundles. The headline change is simple: production persistence no longer depen
A prompt buried in an f-string is technical debt. You can't test it. You can't save it to a file. You can't hand it to a colleague and say "here's the extraction logic." The moment you hard-code a prompt into a string, you've welded your application logic to a single, frozen instruction — one that w
TL;DR I no longer recommend Railway for serious production workloads after its recent pattern of incidents. Fly.io is not simpler, but it is one of the first alternatives worth evaluating if you want more control over infrastructure and blast radius. Even if Fly.io is not your final choice, I think
Bilingual post · Post bilíngue Jump to: English · Português Delphi projects rarely compile as a single flat source file. {$IFDEF}, {$DEFINE}, {$INCLUDE}, and friends select platform code, toggle debug logging, and stitch units together. CrabPascal ships a line-oriented preprocessor (since v1.5.2, re
Key Takeaways Use RAG for knowledge retrieval, changing data, and rapid iteration. Use fine-tuning for style, format, narrow classification, and cost at scale. Start with RAG — 70% of production problems don't need fine-tuning. Fine-tuned Qwen2.5-7B reached 88% accuracy on a proprietary classificati
