https://cuigh.com/tags/ai/ Recent content in AI on AI Pulse Hugo en-US Mon, 18 May 2026 12:27:32 +0800 https://cuigh.com/posts/ai-video-tools-from-prompts-to-workflows/ Mon, 18 May 2026 12:27:32 +0800 https://cuigh.com/posts/ai-video-tools-from-prompts-to-workflows/ Vivago Video Agent, Velo 2.0, Open-Generative-AI, Pixelle-Video, and hyperframes all point to the same shift: AI video is moving from prompt tricks to production workflows. https://cuigh.com/posts/companies-ai-psychosis/ Sun, 17 May 2026 10:30:56 +0800 https://cuigh.com/posts/companies-ai-psychosis/ A Hacker News thread about AI psychosis became a useful warning: companies are overestimating AI agents, underestimating verification costs, and confusing demos with reliable workflows. https://cuigh.com/posts/claude-code-skills-ai-coding-assets/ Sat, 16 May 2026 10:36:46 +0800 https://cuigh.com/posts/claude-code-skills-ai-coding-assets/ Claude Code Skills, CLAUDE.md files, Codex Skills, and agent workflow repos are all showing up on GitHub Trending. The real story is not longer prompts, but reusable engineering judgment packaged as assets for AI agents. https://cuigh.com/posts/openai-codex-mobile-ai-coding-agent-remote-control/ Fri, 15 May 2026 12:38:48 +0800 https://cuigh.com/posts/openai-codex-mobile-ai-coding-agent-remote-control/ OpenAI has brought Codex into the ChatGPT mobile app. The real story is not coding on a phone, but using the phone as a lightweight control surface for long-running AI coding agents. https://cuigh.com/posts/ai-media-workflow-2026-agents/ Thu, 14 May 2026 15:06:35 +0800 https://cuigh.com/posts/ai-media-workflow-2026-agents/ A practical look at how tools like RankSpot, Velo, Postiz, Pixelle-Video, and ppt-master point toward a new AI-powered creator workflow, where agents handle the repetitive execution while humans keep the judgment. https://cuigh.com/posts/stop-asking-ai-for-markdown-ask-for-html/ Wed, 13 May 2026 15:14:24 +0800 https://cuigh.com/posts/stop-asking-ai-for-markdown-ask-for-html/ From Karpathy's suggestion to ask LLMs for HTML output to products like display.dev and ppt-master, this essay argues that AI output is moving from text answers toward pages, interfaces, and lightweight deliverables. https://cuigh.com/posts/ai-life-experiments-small-annoyances/ Tue, 12 May 2026 15:50:02 +0800 https://cuigh.com/posts/ai-life-experiments-small-annoyances/ From school newsletters and nighttime noise to piano practice, this essay argues that the most practical AI automations for everyday people start with small recurring annoyances, not grand all-purpose assistants. https://cuigh.com/posts/personal-agent-infrastructure-mcp-memory-skills/ Mon, 11 May 2026 15:57:18 +0800 https://cuigh.com/posts/personal-agent-infrastructure-mcp-memory-skills/ From the rise of Skills and Memory projects on GitHub to GBrain's MCP thin client, Claude Managed Agents memory, and tool-routing products on Product Hunt, this essay argues that the next generation of personal AI agents needs three layers of infrastructure: rules, memory, and tool access. https://cuigh.com/posts/cheap-local-ai-tools-are-starting-to-replace-closed-source/ Sat, 09 May 2026 14:17:35 +0800 https://cuigh.com/posts/cheap-local-ai-tools-are-starting-to-replace-closed-source/ From DeepSeek-TUI and local-deep-research to 9router, this essay looks at why local AI tools are moving from hacker toys to serious alternatives for individuals and small teams. https://cuigh.com/posts/when-ai-starts-spending-money/ Thu, 07 May 2026 14:54:07 +0800 https://cuigh.com/posts/when-ai-starts-spending-money/ From pay.sh to Cloudflare's agent-execution examples, this essay looks at a new shift in AI: systems are no longer just advising humans. They are starting to discover services, call APIs, and spend money on our behalf. https://cuigh.com/posts/voice-is-eating-the-prompt/ Wed, 06 May 2026 14:42:42 +0800 https://cuigh.com/posts/voice-is-eating-the-prompt/ From Sam Altman's view on voice interfaces to Velo 2.0, GLM-5V-Turbo, and GBrain, this essay argues that prompting is not disappearing, but it is slowly moving from the front door to the backend. https://cuigh.com/posts/stop-asking-ai-for-pictures-ask-for-svg-instead/ Wed, 29 Apr 2026 21:45:25 +0800 https://cuigh.com/posts/stop-asking-ai-for-pictures-ask-for-svg-instead/ Starting from Zara Zhang’s argument that AI should generate more SVG and fewer static images, this piece looks at why editable, reusable structured output often creates more workflow value than a flashy visual. https://cuigh.com/posts/business-agents-that-make-money-stay-close-to-sales-and-launch/ Tue, 28 Apr 2026 09:09:52 +0800 https://cuigh.com/posts/business-agents-that-make-money-stay-close-to-sales-and-launch/ Using Orange Slice, Jet AI Agents, and Waitlister as examples, this piece argues that the business agents most likely to survive are the ones closest to sales, launches, acquisition, and conversion. https://cuigh.com/posts/agents-are-not-the-bottleneck-work-is/ Fri, 24 Apr 2026 12:22:22 +0800 https://cuigh.com/posts/agents-are-not-the-bottleneck-work-is/ Starting from a sharp line by Zara Zhang, this piece connects Workspace Agents, Teams Agent, Zed Parallel Agents, and Every's employee-agent experiment to argue that what is scarce today is not agents. It is work that is structured, permissioned, and reviewable enough to hand to them. https://cuigh.com/posts/why-new-model-releases-matter-less-than-value/ Mon, 20 Apr 2026 09:04:34 +0800 https://cuigh.com/posts/why-new-model-releases-matter-less-than-value/ Using the Hacker News debate around Opus 4.6 and 4.7 as a starting point, this piece looks at why developers now care less about new model releases and more about whether the upgrade is actually worth it. https://cuigh.com/posts/why-ai-tools-are-rushing-to-mac-desktop/ Sat, 18 Apr 2026 23:01:19 +0800 https://cuigh.com/posts/why-ai-tools-are-rushing-to-mac-desktop/ A trend analysis and product review of Gemini on Mac, Ollama v0.19, and Google AI Edge Gallery, and why low-friction experience matters more than raw model strength.