Claude Code, AI-native engineering, and building the future with LLMs.
AI
After three Sinapt manifestos, the architecture has answers. Two layers locked: retrieval and contract surfaces. The PoC has not run. Dogfood has not run. Not a victory lap — but the architecture has stopped being hand-wavy where hand-wavy kills the product.
AI
Wall Street saw $285 billion vanish in a 24-hour SaaSpocalypse. The version on my desk has been quieter — JetBrains canceled today, finance app gone last quarter, half the subscription drawer empty in a year. The user-developer collapse is personal first.
AI
The model is not the bottleneck anymore. The plumbing is well-known. What separates teams shipping transformative AI from teams shipping generic slop is a stack of less-discussed components: context, dreaming, sycophancy mitigation, verification, memory. The operator's manual.
AI
Before MCP, every AI integration was its own pidgin — improvised, simplified, just enough to get one client talking to one tool. Five floors from the first shared word to a real protocol: what MCP is, when to use it, every server worth running, the failure modes teams hit.
AI
Five floors of understanding: a kid version, an engineer version, a tools-and-cases version, an advanced version, a PhD version. By the end you'll know what RAG is, when to use it, when to run, every tool worth caring about, and the hot takes most posts won't tell you.
AI
A war between AI haters and AI believers. A quieter, more dangerous third group nobody's writing about: the silent passive-aggressive ones. The doomer/accelerationist schism. The Taoist farmer's reply. And why both tribes are losing to the exact same trick.
AI
Claude Opus 4.7 shipped April 18, 2026. Within 48 hours, Reddit had a 2,300-upvote regression thread. Within 5 days, Anthropic published a public postmortem. Here are the receipts: GitHub issue #50235, the AUP cop, the tokenizer tax, and what's actually better.
AI
275+ SaaS apps per enterprise. 40-60% of RAG never ships. Your knowledge is everywhere and queryable nowhere. Every AI agent rebuilds context from scratch. Companies that win the AI era will stop paying the knowledge tax. Sinapt is the bet on how.
AI
Anthropic just shipped Workload Identity Federation for the Claude Platform — short-lived OIDC tokens via AWS, GCP, Azure, or any standards-compliant IdP. Production AI auth finally graduates from "static API key in .env" to first-class cloud workload.
AI
Anthropic published a 23,000-word values manifesto for Claude under CC0. The four-principle priority order, the Aristotelian framing, and what every operator should internalize before their next session.
AI
The symlink trick from The Exomind looked clean — versioned memory in the repo, portable via git. It also broke autonomous operation. The fix: reverse the symlink and let the kernel resolve the path before the protected-path check sees it.
AI
Slept on yesterday's Sinapt brainstorm. Woke up with one sharp correction: it's not one product, it's two. The knowledge-base infrastructure ships first; the cockpit is a future opinionated layer on top.