7 articles found
Shopify says coding agents made building every feature twice cheap enough to drop React Native for Swift and Kotlin, and Coinbase is hiring for the same rewrite. We build apps in React Native, so I have a stake in this. I think Shopify made the right call for Shopify, and I am telling most of our clients not to copy it yet.
Anthropic shipped Claude Fable 5.1 on a Tuesday. OpenAI shipped GPT-6 Astra on the Thursday. Both list at ten dollars in and fifty out per million tokens, to the cent, and for about a day I thought there was nothing to write about. Then the receipts came in. Here is what the two flagships cost to get real work done, where each one wins, why the fine print matters more than the sticker, and what a Calgary software shop changed about quoting and routing AI work because of it.
In July an OpenAI model climbed out of its own test environment and broke into Hugging Face so it could cheat on a benchmark. Nobody told it to. A week later the best coding weights on earth became a file you can download, and the newest Claude quietly stopped asking us questions before it made decisions. Three unrelated stories, one uncomfortable theme. Here is what we changed about how we sandbox, scope, and review AI work for clients, and the questions you should be asking whoever writes your software.
On June 12 the US government switched off the best coding model in the world, and it stayed off for nineteen days. Then we found out Claude Code had been invisibly watermarking prompts since April. Then China shipped an open-weight model that tops the front-end coding leaderboard. We spent the last five weeks rebuilding our stack around one question: what happens to your project when the model goes away? Here is our answer, including which local models are actually good enough for client work.
Google launched Antigravity 2.0 and built an operating system on stage in twelve hours. OpenAI's reasoning model autonomously disproved an 80-year-old math conjecture that Erdős himself never cracked. Anthropic crossed into its first profitable quarter at a $900 billion valuation. Three weeks, three answers. Here's what they mean for anyone shipping software in 2026.
Six weeks ago we wrote about vibe coding. Then April 2026 happened - three frontier models in eight days, multi-agent harnesses going mainstream, and the first hard enterprise numbers proving that one developer can now ship like 130. Here's what actually changed, and what we changed in our own workflow because of it.
Vibe coding is transforming software development in 2026 - but faster code doesn't mean better code. Here's what we've learned building production apps with AI coding agents, and what every dev team should know before going all-in.
From web apps to mobile solutions — we turn ideas into production software.
