--- title: "AI Works Hard, But It Has No Idea What You're Doing: Practical Reflections from Multi-Agent to Vibe Coding" description: "Introduction: The more I use AI, the more I feel it doesn't really understand what you're doing. This isn't me being pessimistic, but an honest takeaway after getting my hands dirty for a while. When you give it a complex task, the quality of its output is inherently capped by its limited context. It's fast and diligent, but it doesn't \"understand\" what you're doing. In this post, I want to discuss three things: the essence of Multi-Agent, the hidden aspects of Vibe Coding..." canonical_url: "https://blog.markkulab.net/en/post/multi-agent-vibe-coding-practical-reflection" author: "Mark Ku" author_url: "https://blog.markkulab.net/en/author/mark-ku" site: "Mark Ku's Tech Notes" date_published: "2026-04-04 15:32:52 +0800" category: "AI" language: "en" license: "CC BY 4.0" license_url: "https://creativecommons.org/licenses/by/4.0/" attribution: "when reusing or quoting, credit the author and link back to the original" --- # AI Works Hard, But It Has No Idea What You're Doing: Practical Reflections from Multi-Agent to Vibe Coding ## Introduction The more I use AI, the more I feel like it doesn't really understand what you're doing. This isn't me being pessimistic, but an honest takeaway after getting my hands dirty with it for a while. When you give it a complex task, the quality of its output is capped by its limited context. It's fast and diligent, but it doesn't *understand* what you're doing. In this post, I want to talk about three things: the nature of Multi-Agent systems, the hidden costs of Vibe Coding, and which skills are becoming more valuable in an era accelerated by AI. ## Multi-Agent: Orchestrated Automation, Not Autonomous Driving My own blog runs on a LangGraph Multi-Agent Pipeline that chains everything together, from research and writing to cover art, voiceovers, and video. After playing with it for a while, my conclusion is: > At its core, Multi-Agent is still "human-orchestrated automation." It adds logic to the call-and-response flow, allows for repeatable execution, and lets you use agents with different personas to help, but it's a long way from being fully autonomous. A 2026 CIO magazine article directly stated that "true multi-agent collaboration isn't feasible," as agents will exhibit issues like: - ๐Ÿ”„ Ignoring instructions, repeating work - ๐Ÿ“‰ Delegation failures, planning paralysis - โš™๏ธ Coordination overhead becoming a bottleneck itself These problems are almost identical to management issues in human organizations. The bottleneck isn't the capability of a single model, but the complexity of coordination. ![AI article writing tool interface with a complex multi-agent workflow diagram](https://blog.markkulab.net/content/markku/posts/multi-agent-vibe-coding-practical-reflection/images/667429162_26583505644577677_1599853590001067562_n.jpg) ## Vibe Coding: Accelerating Development, and Also Chaos The concept of Vibe Coding is appealing: describe your needs in natural language, and the AI directly generates runnable code. But from my observation, this can create a host of problems: ### ๐Ÿ“Š What the Data Says

Metric

Data

Security vulnerability rate in AI-generated code

53% (2.74 times higher than human-written code)

Number of vulnerabilities found in vibe-coded applications

Over 2,000 vulnerabilities and 400+ exposed secrets in 5,600 applications

Startup products needing partial or complete rewrites

~8,000 companies, with an estimated cost of $0.4โ€“4 billion

Maintenance cost (Year 2)

4 times that of traditional development

Time developers spend per week fixing AI errors

11 hours on average

2026 has even been dubbed the "Year of Technical Debt." ### ๐Ÿงจ Four Hidden Dangers of Vibe Coding **1. Digital Silos** Everyone uses AI to quickly generate their own systems. Without a shared architectural language, these systems can't be integrated or reused, eventually becoming isolated islands. **2. Unmaintainable Code** Most Vibe Coding projects hit a "Spaghetti Point" around the third month: new features break existing ones, and development velocity grinds to a halt. A deeper problem is the "accountability gap": when a bug is found six months later, no one can explain why the code was written that way, because no human actually made that decision. **3. Hard-to-Trace Logic** AI-generated code might seem to work, but it lacks a coherent design intent. No one can answer "Why was this pattern chosen?" or "Why use this data structure?" because the answer is "The AI chose it." **4. Security Vulnerabilities** AI models are trained on vast amounts of open-source code, which includes many vulnerabilities. The AI will naturally reproduce these issues. For users without a security mindset, they have no idea what they're deploying. ### ๐Ÿ’ก The Most Easily Overlooked Fact Many people mistakenly believe "With AI, everything can be accelerated." But the most time-consuming parts of software development have never been the act of writing code itself. They are: - Clarifying ambiguous and uncertain requirements - Managing system complexity - Long-term maintenance and evolution AI currently offers little help with these tasks. Letting someone unfamiliar with the domain use AI to speed up development usually doesn't accelerate output; it accelerates chaos. ## What is AI Truly Replacing? In my experience, what AI can effectively replace right now are repetitive tasks, and the impact of that is already huge: ``` AI ๆ“…้•ท AI ไธๆ“…้•ท โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€ โœ… ้‡่ค‡ๆ€ง็š„ CRUD ้–‹็™ผ โŒ ่ค‡้›œ็ณป็ตฑ็š„้™ค้Œฏ่ˆ‡ๆžถๆง‹ๅ–ๆจ โœ… ๆจฃๆฟ็จ‹ๅผ็ขผ็”Ÿๆˆ โŒ ๅˆ†ๆ•ฃๅผ็ณป็ตฑ็š„ๆนง็พ่กŒ็‚บๅˆคๆ–ท โœ… ๆ–‡ไปถ็ฟป่ญฏใ€ๆ‘˜่ฆใ€ๆ ผๅผ่ฝ‰ๆ› โŒ ๅฐๆŠ—ๅผๅฎ‰ๅ…จๆ€็ถญ โœ… ๆธฌ่ฉฆๆกˆไพ‹็”Ÿๆˆ โŒ ่ทจๅœ˜้šŠๆบ้€š่ˆ‡้œ€ๆฑ‚้‡ๆธ… โœ… ่ณ‡ๆ–™ๆ•ด็†่ˆ‡ๅˆ†ๆž โŒ ๅ•†ๆฅญ้‚่ผฏ็š„ๅ–ๆจ่ˆ‡ๅฆฅๅ” ``` A 2026 McKinsey report points out an interesting phenomenon: after AI coding assistants became widespread, the demand for software developers actually increased by 34%, but the job structure changed dramatically. Companies no longer need a 10-person team to write CRUD applications. Instead, they need 1-2 senior engineers to handle architectural design, review AI output, and manage the gray areas where AI can't make a judgment. ## Conclusion: Execution is a Commodity, Judgment is the Moat My feeling is that current AI isn't powerful enough to replace people, but at most, it can replace repetitive work. But even that alone is incredibly powerful and has already impacted many people's jobs. Value is shifting: - **Decreasing** โ†’ The ability to write code and produce output quickly (Execution) - **Increasing** โ†’ The ability to clarify ambiguous requirements, manage complexity, and ensure maintainability (Judgment) For engineers, the best investment right now isn't learning more AI tools, but strengthening the skills that AI can't replicate: architectural thinking, system design, requirements analysis, and a commitment to quality. AI is a powerful co-pilot, but you still need to keep your hands on the steering wheel. ## References - [2026 Predictions: It's the Year of Technical Debt (Thanks to Vibe-Coding)](https://www.salesforceben.com/2026-predictions-its-the-year-of-technical-debt-thanks-to-vibe-coding/) - [AI Technical Debt: How Vibe Coding Increases TCO](https://www.baytechconsulting.com/blog/ai-technical-debt-how-vibe-coding-increases-tco-and-how-to-fix-it) - [True multi-agent collaboration doesn't work | CIO](https://www.cio.com/article/4143420/true-multi-agent-collaboration-doesnt-work.html) - [53% of AI Code Has Security Holes](https://www.getautonoma.com/blog/vibe-coding-security-risks) - [Will AI Replace Software Developers? The Truth About AI in 2026](https://www.dreamztech.com/blog/will-ai-replace-software-developers-2026/) - [Vibe Coding Is Over. What Comes Next Is Much Harder.](https://medium.com/@ahmed.hafdi.contact/vibe-coding-is-over-what-comes-next-is-much-harder-9fe95b509850) - [The Reality of Vibe Coding: AI Agents and the Security Debt Crisis](https://towardsdatascience.com/the-reality-of-vibe-coding-ai-agents-and-the-security-debt-crisis/) --- ## About this article and its author Originally published on [Mark Ku's Tech Notes](https://blog.markkulab.net/en/post/multi-agent-vibe-coding-practical-reflection) License: [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) โ€” when reusing or quoting, credit the author and link back to the original ### About the author **[Mark Ku](https://blog.markkulab.net/en/author/mark-ku)** โ€” Software Solution Provider - 10+ years senior software engineer, now an AI Builder - Focused on large-platform architecture โ€” North-American e-commerce, AI SaaS subscription billing - Combining AI Agents and automation to build evolvable product foundations ### Free tools built by the author All of these are free to use: - [Free PDF Sign Tool](https://blog.markkulab.net/en/tools/pdf-sign): Online PDF sign tool โ€” draw, type, or upload a signature, then drag, resize, and download. Everything runs in your browser; nothing is uploaded. - [VS Code Refactory](https://blog.markkulab.net/en/tools/refactory): Refactory is a VS Code refactoring extension: 34 actions plus a 37-rule code-smell inspection layer with a Code Health dashboard, across 18 languages, backed by 534 tests. It learns your repo's conventions: where interfaces live, where DI is registered, whether 'use client' belongs. It ranks files by git churn ร— complexity so you know what to fix first, and hands any smell to the Claude Code already on your machine. Free to use, and your source never leaves your computer. - [DB-Kit Database Manager](https://blog.markkulab.net/en/tools/db-kit): DB-Kit is a lightweight, cross-platform database manager built with Tauri + Rust + React. Manage MySQL, MariaDB, PostgreSQL, SQL Server, Oracle, SQLite, MongoDB, Redis, Kafka, Elasticsearch and RabbitMQ from one consistent interface: passwords encrypted in the OS keychain, SSH tunnels, full CRUD, a visual query builder, stacked multi-statement result sets, cross-connection data transfer and compare/sync, Excel / CSV import & export, visualized execution plans, ER diagrams, scheduled backups, SQL stress testing with p50โ€“p99 latency percentiles, a 15-rule SQL review engine, Kafka message browsing with monitoring & alerts, a bilingual UI (Traditional Chinese / English), a built-in AI assistant (natural-language SQL, AI review and tuning advice) and the dbk CLI. Free and open source (MIT), with installers for Windows, macOS and Linux. - [VS Code Super Mermaid](https://blog.markkulab.net/en/tools/super-mermaid): Super Mermaid is a VS Code extension for beautiful Mermaid diagrams out of the box: auto-colored live preview, mouse pan & zoom, high-res PNG / SVG export, 21 templates and multiple themes. Free and open source (MIT). - [React Super Mermaid](https://blog.markkulab.net/en/tools/react-super-mermaid): react-super-mermaid is an open-source React component library: render beautiful Mermaid diagrams with a single , with built-in colorful / sketch themes, pan & zoom, in-diagram search, and high-res SVG / PNG export. Lightweight, SSR-safe, fully typed. Free and open source (MIT). - [Jira / Confluence Super Mermaid](https://blog.markkulab.net/en/tools/jira-super-mermaid): An Atlassian Forge app: write Mermaid syntax directly inside a Jira issue or a Confluence page and get flowcharts, sequence diagrams, state machines and Gantt charts. 11 diagram types, SVG / PNG export, light and dark themes, full CJK support. Runs on Atlassian: your diagrams live in your own site and the app calls no third-party service. Free, coming soon to the Atlassian Marketplace. - [Mermaid Live Preview](https://blog.markkulab.net/en/tools/mermaid-preview): Write Mermaid in your browser, see it render instantly, and share the whole diagram as a single link. No sign-up, nothing uploaded to a server, and mermaid.live share links work as-is. - [React Intl Phone Number](https://blog.markkulab.net/en/tools/react-intl-phone-number): react-intl-phone-number is an open-source React component: framework-agnostic and antd-free, with E.164 in/out, a searchable flag / country-code dropdown, configurable validation levels (strict / mobile-strict / loose), themeable CSS, and i18n โ€” phone logic powered by google-libphonenumber. Lightweight and fully typed. Free and open source (MIT). - [Uptime Kuma Cluster](https://blog.markkulab.net/en/tools/uptime-kuma-cluster): Turn single-node Uptime Kuma into a highly available cluster: OpenResty + Lua smart load balancing, shared MariaDB state, health checks and automatic failover, plus cluster-management REST APIs. One Docker Compose command to start. Free and open source (MIT). - [Special Education](https://blog.markkulab.net/en/education): Learning materials crafted for special education students ### Daily podcasts - [Mark's Tech Insights โ€” Daily AI News](https://blog.markkulab.net/en/category/tech-news): Daily curated AI and tech trends. Catch the latest developments via audio summaries โ€” covering AI applications, software architecture, DevOps, and engineering practice. โ€” RSS: https://blog.markkulab.net/feed.xml - [AI่‚กๅธ‚่ฆ่Š](https://blog.markkulab.net/en/category/ai-stock-chat): Every trading day, an AI-analyzed take on the Taiwan stock market, delivered as a two-host conversation covering the session and the next-day outlook. โ€” RSS: https://blog.markkulab.net/ai-stock-chat/feed.xml - [้–‹ๆบๅฅฝ็‰ฉ้€ฑๅ ฑ](https://blog.markkulab.net/en/category/open-source-weekly): A weekly two-host pick of free open-source tools surfaced from real Hacker News, GitHub, and Reddit buzz โ€” what pain they solve and the fastest way to get started. โ€” RSS: https://blog.markkulab.net/open-source-weekly/feed.xml ### Newsletter [Subscribe to the newsletter](https://blog.markkulab.net/en/subscribe) โ€” Be the first to know about new posts. No spam, unsubscribe anytime.