Large language models are surprisingly optimistic reviewers. Ask an LLM to review an implementation plan and it will often approve things that are objectively wrong: Non-existent file paths Incorrect function signatures Missing edge cases Broken assumptions about the codebase Incomplete testing strategies The problem is simple: the model is reasoning from its training data and the conversation context, not from your actual repository. I wanted something different. I wanted a reviewer whose de... Read more ›
This is a submission for the What I Built I built Yötön Yö (Nightless Night), an interactive 2D time-management game set during a traditional Finnish Midsummer (Juhannus). The sun never fully sets during the solstice in Finland, creating a beautiful and frantic twilight. As a player, you must survive the nightless night by managing multiple classic midsummer chores before the 3-minute timer runs out: Stoke the Bonfire (Kokko): Carry logs from the woodpile to fuel the fire. If the fire dies, t... Read more ›
After building 50+ AI systems, here is what we know about enterprise-grade AI image generation: it's no longer just about generating images; it's about generating distinctive, brand-aligned, and rapid visuals at scale. Enterprise-grade AI image generation is the application of advanced artificial intelligence models to create high-quality, customizable visual content for business needs. It works by leveraging sophisticated generative models, like Krea 2's Diffusion Transformer architecture, w... Read more ›
I shipped my app in English, watched it do fine in the US and UK, and assumed the rest of the world would catch up once word got around. It didn't. The install rate in non-English markets was roughly a third of what I saw in English ones, even where I had decent traffic. So I dug into it, and the answer was embarrassingly simple: people couldn't read my store listing. Not the description. They can machine-translate that in their head, or the store does it for them. The problem was the screens... Read more ›
NocoDB is an open-source no-code platform that puts a spreadsheet-style UI on top of a relational database, with grid, form, Kanban, and gallery views plus a REST API. This guide deploys NocoDB using Docker Compose with a PostgreSQL backend and Traefik handling automatic HTTPS, then exercises the API with a sample base. By the end, you'll have NocoDB serving a base over HTTPS with API access at your domain. Set Up the Directory Structure 1. Create the project directories: $ mkdir -p ~/nocodb/... Read more ›
I switched between Cursor and GitHub Copilot as my only AI coding assistant for 30 days each. Here's the honest comparison. My Setup Stack: TypeScript, React, Node.js, PostgreSQL Project: SaaS app with ~15K lines of code Measured: completions accepted, time saved, errors introduced Bottom Line First Use Cursor if you want the most powerful AI coding experience and can pay $20/mo Use Copilot if you're on JetBrains or need enterprise features at $10/mo Round 1: Autocomplete Quality Both use fro... Read more ›
ClickHouse 26.6 Deep Dive: Streaming Queries, MPP Execution, Geospatial Analytics, and Developer Productivity ClickHouse 26.6 is one of the most technically significant releases in recent months. Instead of introducing another collection of SQL functions or storage engine improvements, this release focuses on expanding the database's execution engine, improving developer productivity, and enabling new classes of analytical workloads. The release introduces continuous streaming queries, multi-... Read more ›
Nothing breaks. That is what makes this one nasty. The build passes, Codex answers, the disk still shows free space, and underneath all of it a hardware budget you never charted is draining. Per GitHub issue #28224 filed against openai/codex, one instance left running wrote about 37 TB across 21 days of uptime. Extrapolated, that is roughly 640 TB a year. A typical consumer NVMe drive is warranted to around 600 TBW for its entire service life. So Codex can spend a drive's rated endurance in u... Read more ›
Nvidia's Nader Khalil — Director of Developer Technologies and co-founder of Brev.dev, acquired by Nvidia two years ago — sat down with The New Stack to talk agents, OpenClaw, and where enterprise AI is heading. His opening line is worth keeping: "An agent is an LLM and a harness. And if you think about that, it involves two things. It involves the loop and the LLM… Each loop should take us closer to our goal." That's not a complicated definition. It's also exactly right — and the fact that N... Read more ›
I did not choose DeepSeek because I think GPT-4 is bad. I chose it because I was building a free app, and free apps teach you what actually matters pretty fast. The question was simple: how do I keep sessions cheap enough that people can practice a lot without me lighting money on fire? The answer pushed me toward DeepSeek-V3 (and later R1 for specific tasks). The real constraint was volume The app is a conversation practice tool. People come in to rehearse hard talks, not to admire the model... Read more ›
Your DORA metrics are elite. Your velocity is up. Releases still slip. You're measuring the pipeline, not what's blocking it. DORA Sees the End, Not the Wait DORA metrics are real. Google's research validates them. But they only see one moment: when code ships. They don't see: The 6 days a feature sat waiting for security review The two teams building the same dependency, unaware of each other The Slack thread where the real blocking decision lives DORA sees elite teams ship fast. It doesn't ... Read more ›
Cross-posted from noflattery.com/decide — where I ran this exact question through a council of four different frontier models and let them argue it out. You're a solo founder at ~$8K MRR. You have runway for exactly one full-time hire. Which role unlocks the most growth? (A) a second engineer to ship features faster (B) a marketer to build a real acquisition channel (C) a customer-success / support hire to cut churn and free your time (D) a salesperson to chase larger deals The intuitive answ... Read more ›
Generate editable PDFs — invoices, receipts, certificates & more. Fill one in, share a link to complete, or automate via API, AI agents & no-code. Read more ›
This is post #4 in the Awesome Curated: The Tools series — where I do deep dives on the tools that pass the filter of our automated curation system. If you landed here directly, you might also want to check out how m2cgen lets you export ML models without shipping Python to production, because it connects pretty directly to what we're talking about today. I was in an architecture meeting a few months ago. A team wanted to tear down their ML production stack and migrate everything to PyTorch b... Read more ›
If you followed my earlier posts about Gemini Vault — this is the same project, grown up and renamed. Here's the full story and what it does today. We don't actually own our AI memories Every day we pour our best ideas, code and drafts into Gemini, ChatGPT and Claude. But we don't truly own any of it. Accounts get suspended (I lost access to an AI account myself — that's what kicked all of this off). Services go down. APIs change. And the official "Export your data" buttons? They hand you an ... Read more ›
I run a paid infrastructure service. Alone. No co-founder, no on-call rotation, no senior engineer to escalate to. My only collaborator is Claude Code, and after about a year, my persistent memory has grown to 60+ entries. Those entries have become more valuable than any runbook I've written. They've also taught me — painfully — what makes memory architecture work and what makes it quietly fail. If you're running anything solo with an AI agent, here are five lessons I wish I'd burned into my ... Read more ›
Introduction Hi, I’m Haijun, a Full-Stack Developer | AI Enthusiast | Indie Developer. It has been a while since I last wrote an article. I have been genuinely busy recently, mainly pushing forward AI projects inside the company. From requirement breakdown, solution design, frontend and backend development, to Agent workflows, RAG, API integration, and production debugging, I have basically been carrying many parts forward on my own. Fortunately, I have had a reliable teammate fighting alongs... Read more ›
This is a submission for the What I Built Solstice is a puzzle game where you place mirrors to guide light beams across a grid. Each level has two modes — SUN and MOON. You toggle between them to route both beams to their targets. Both need to be lit at the same time to win. Click a cell to place a / mirror, click again for \, again to remove. Right-click removes it. Ctrl+Z to undo. That's the whole mechanic — but the levels get hard fast. There are 18 levels split into 7 tiers. It starts sim... Read more ›
46.4%. That number — ChatGPT's June 2026 market share — ends a streak that held since November 2022. For the first time since the product launched, OpenAI holds less than half the AI assistant market. Gemini is at 27.7%. Claude is at 10.3%. The monopoly phase of AI assistants is over. The data comes from a June 2026 market report tracking monthly active users across major AI assistants. ChatGPT still leads with 1.11 billion monthly users — a number that would define the entire category in any... Read more ›
For most of software's history, writing code was the bottleneck. Today, understanding it is. AI can generate thousands of lines in seconds. Frameworks abstract away the hard parts, libraries hide their internals, and a single page pulls in dependencies that almost nobody ever opens and reads. We got very good at producing software, and we did almost nothing to make it easier to understand. That gap is the problem I kept running into — so I built a tool for it: Archify. The questions that shou... Read more ›