Vibez.now blog https://vibez.now/blog Notes on building software with AI agents: autonomy, approvals, long runs, and what it costs. en Build Your Black Friday Page This Week, Not on Thanksgiving Night https://vibez.now/blog/build-your-holiday-page-before-black-friday https://vibez.now/blog/build-your-holiday-page-before-black-friday Sun, 01 Nov 2026 12:00:00 GMT Small business A practical plan for small businesses building a Black Friday or Cyber Monday page with an AI app builder in November 2026: what to build, when to stop changing it, and the checks that matter before the traffic arrives. The Agent Can Use Your Stripe Key. It Should Never Be Able to Read It. https://vibez.now/blog/ai-agent-api-keys-secrets https://vibez.now/blog/ai-agent-api-keys-secrets Sun, 01 Nov 2026 12:00:00 GMT Security How AI app builders should handle API keys and secrets: why an agent should never see a secret's value, the difference between server and browser keys, and why the build output is what gets scanned. The Web Page Told Your Agent What to Do Next https://vibez.now/blog/prompt-injection-ai-coding-agents https://vibez.now/blog/prompt-injection-ai-coding-agents Sun, 01 Nov 2026 12:00:00 GMT AI Safety & Workflows Prompt injection is the security problem every AI coding agent shares. Why detection does not solve it, and how gating actions instead of reads keeps an injected instruction from reaching anything that matters. Before 2027: Who Owns the App the AI Built? https://vibez.now/blog/who-owns-your-ai-built-app https://vibez.now/blog/who-owns-your-ai-built-app Sun, 01 Nov 2026 12:00:00 GMT Teams A year-end checklist for AI-built apps that turned into real tools: code ownership, history, access, secrets and handover, so the app your team relies on still has an owner in 2027. Nobody Reads the Approval Dialog https://vibez.now/blog/nobody-reads-the-approval-dialog https://vibez.now/blog/nobody-reads-the-approval-dialog Mon, 07 Sep 2026 12:00:00 GMT AI Safety & Workflows Approval fatigue defeats most human-in-the-loop AI systems. The fix is not better warnings — it is asking less often, and defaulting closed on the things nobody has reasoned about. One of These Models Costs 5× More Than the Other. Most Tasks Can't Tell. https://vibez.now/blog/which-model-should-build-your-app https://vibez.now/blog/which-model-should-build-your-app Sat, 05 Sep 2026 12:00:00 GMT Cost & models Opus, Sonnet and Haiku differ by 5× on input and 5× on output. Here is how to think about which one should run a given turn, and why the picker and the price list have to be the same table. Your AI Agent Says the App Works. Here's the Proof It Doesn't. https://vibez.now/blog/your-ai-agent-says-it-works https://vibez.now/blog/your-ai-agent-says-it-works Fri, 04 Sep 2026 12:00:00 GMT Testing Compiling is not working. A React app that throws on mount still answers 200 with an empty root div — so build checks and status-code sweeps both pass while the user stares at a white screen. We Let an AI Agent Build for Ten Hours Unattended https://vibez.now/blog/ten-hour-ai-agent-run https://vibez.now/blog/ten-hour-ai-agent-run Wed, 02 Sep 2026 12:00:00 GMT Autonomous agents Long-running AI coding agents can work for hours without you. Here is what actually governs one: iteration limits, wall-clock deadlines, spend ceilings, and the checkpoint that decides when it is done. What is Human-in-the-Loop (HITL) AI? https://vibez.now/blog/what-is-human-in-the-loop-ai https://vibez.now/blog/what-is-human-in-the-loop-ai Tue, 16 Jun 2026 12:00:00 GMT AI Safety & Workflows Learn what Human-in-the-Loop (HITL) means for AI agents, why it matters for safety, and how to implement approval workflows with Vibez.now.