Reinventing the LibGuide with Natural Language Programming

AI vibe coding opens up new possibilities for library resources

Posted on January 21, 2026 in Blog Posts

A few months ago, I decided it was time to learn how to vibe code. I didn’t know exactly what I’d build, but I wanted to try. “Vibe coding” is the latest buzz in Silicon Valley—a term popularized in early 2025 by AI researcher Andrej Karpathy to describe programming by expressing intent in natural language rather than writing syntax line by line.

In practice, it means writing prompts in English—or in your native language—and having a generative AI system translate those ideas into functional code. At first glance, it’s easy to underestimate how revolutionary that is. The ability to describe what you want a website or app to do and then watch an AI agent assemble the code changes the entry point for software creation entirely.

When I first opened Lovable.dev, one of the latest vibe-coding tools, I typed out what I wanted: a personal website that organized my writing, podcast appearances, and talks. I described the general layout, the tone, and even the “feel” I wanted for the site—clean, calm, minimal, like Apple.com if Apple made a personal website. The AI parsed my description, interpreted it as design intent, and began constructing page layouts.  

To my surprise, Lovable’s agent didn’t simply generate static templates. It appeared to draw context from my prompts and online presence, KyleBylin.com. To help it along, I provided several Perplexity Deep Research reports summarizing my past work, so the agent could ground its design choices in actual data about me. The result wasn’t perfect, but it was recognizably mine. It was a first draft of my online identity, assembled through collaboration with an AI partner rather than a web designer or developer.  

That moment reframed my understanding of work itself. It wasn’t just about using AI to automate a task; it was about conversing with AI to create something new. I started thinking about this as vibe work: a mode of human–AI collaboration where the core skill is the ability to articulate your goals, aesthetics, and intuitions clearly enough for an AI system to operationalize them.  

To “vibe,” in this sense, means to collaborate conversationally with a generative AI tool—discussing goals, experimenting with approaches, refining outputs through dialogue, and giving feedback in natural language. The “vibe” captures not just function but feeling. I might say, “Make my website feel like Apple’s—minimal, confident, modern,” and the system translates that metaphor into layout choices, typography, and spacing.

This concept of vibe work represents a shift in how we communicate ideas to technology. It replaces the need for precise syntax with semantic clarity, that is, knowing what you want, why you want it, and being able to describe it meaningfully. It is, in many ways, what IBM describes as a move “from syntax to semantics” in software creation, where natural language becomes the new programming interface.  

---

🔥 Stay up-to-date with LibTech Insights by signing up for our free newsletter. Just one weekly email with our new blog posts, top tech news stories, and other bonus content. Check out some posts from our archive:

• Toward More Critical Futures of AI in Academic Libraries
• How to Find and Fix Hidden Accessibility Issues
• Filling in the Gaps: A Guide for Data Imputation for Libraries

💜 ALA is having its 150th anniversary. Come celebrate with us!

---

From Typing to Talking: How Agents Reshape Roles

I started thinking about how this agentic shift would change traditional roles. For example, I use Wispr Flow, a voice-to-text app that has logged over 455,000 words from me—roughly eight books’ worth of dictation. I no longer type most of my writing; I speak it. The process is fast, fluid, and deeply conversational. It’s also rewiring how I think. When I speak my drafts, my brain structures ideas differently—less linear, more associative. My relationship with text feels more alive, almost like teaching an attentive listener who responds with structure and clarity.  

This style of work is where AI tools shine. They amplify intuition. They make it easier to stay in flow. But they also expose new risks. Wired recently warned that vibe coding “could inherit the chaos of open source in the worst way possible” if AI-generated code is opaque or insecure. That critique is fair. When the machine generates code faster than a human can audit it, quality assurance and security testing must evolve just as quickly.  

Still, the potential outweighs the risk. Once employees can describe workflow improvements in plain language—and AI systems can implement them safely—entire layers of inefficiency could disappear. The rise of the AI automation engineer, a job that focuses on identifying opportunities for automation within organizations, will likely accelerate this transformation. Instead of asking whether AI will replace jobs, we’ll ask how people can direct AI more effectively.  

The real unlock, though, came when I realized I could treat my published work—my essays, websites, and presentations—as a personal knowledge API. Instead of letting AI search the open web for information, I could have it reference my verified website. This flipped the relationship between author and algorithm. I wasn’t just feeding prompts into a black box—I was creating a professional identity that AI could access and reuse across projects.  

Imagine what this could mean for students or early-career professionals. A library student could vibe code a digital archive of research projects. A designer could build a portfolio that automatically updates with new work. A news journalist could create an AI that drafts newsletters based on their verified reporting. Each person could effectively build a self-updating, AI-readable record of their expertise—a living database that both humans and machines can learn from.  

This democratization of creation may prove to be vibe coding’s most profound impact. It lowers the barrier to invention. It allows nonprogrammers to become builders, thinkers, and designers. It encourages people to experiment, iterate, and publish ideas in formats that used to require specialized technical knowledge. This is the shift from telling a computer how to do something to telling it what you want done.  

When I step back and look at the trajectory of these tools—Lovable and Wispr—I see a clear through line: each one teaches you how to think with AI. Learning one tool lowers the barrier to learning the next. The skills compound. The interfaces blend. We’re heading toward an era of AI studios—digital spaces where humans and agents cocreate in real time. Students won’t just study ideas; they’ll prototype them. They’ll create a website or app for the final. Researchers won’t just analyze data; they’ll converse with it. Creativity and productivity will start to converge.  

Vibe coding isn’t just about writing software differently. It’s about reimagining human creativity itself. It’s about describing the world you want to build—and watching an intelligent system start building it with you. 

Traditional LibGuides Meet Live-Coded Websites

In the first year of my librarian career, the instinct to build a LibGuide for AI literacy for my university was nearly automatic. Academic librarians know standard outline: homepage, tabs for databases, a few search engine tips, maybe an embedded YouTube video. LibGuides are reliable and structured, but they’re also static. Once published, updating them feels like doing layout with mittens on. The guide freezes my thinking at one moment in time instead of evolving with the AI age.

Vibe coding, or what could more accurately be called “live-coding,” cracked that open. What if the “guide” wasn’t a fixed page but a live-coded website that could grow alongside the class? Instead of dragging boxes around, I could describe the site I wanted: an interactive guide, not a filing cabinet. A place where students could explore resources, test prompts, and see examples generated in real time.

That idea led me to create to Music2045.com, a website inspired by AI 2027, which is a frontier AI timeline that explores when disruptive societal changes might occur if AI advances accelerate. I also wrote an intentionally provocative essay called “Will Musicians Be Extinct in Ten Years?” The site now acts as a living, AI-native guide rather than static resource list. With live-coding, I could simply say, “Create a timeline that shows how AI might reshape music. Connect each moment to historical events and articles.” The agent produced a site that felt more like a narrative map than a LibGuide.

Music2045.com grew into an open-access book as well. I took an existing book that I published and updated its content with all the recent essays that I had written about music and AI. Together, they form an AI-native guide: part timeline, textbook, and guide. Students can move from a timeline entry into a chapter, from a chapter into an activity, and back into new events they help surface.

This shift changes what librarians can prototype. LibGuides optimize for stability; live-coded sites optimize for experimentation and narrative. I can spin up alternate versions for different audiences or build environments where students can learn subjects in brand new and interactive ways.

There’s a deeper shift underneath. A LibGuide assumes knowledge is something we present. A live-coded site assumes knowledge is something we do, together, in real-time. When I vibe code, I’m shaping not just content but workflow: how to move from curiosity to context, from context to critique, from critique to creation. Since the site is malleable, it can evolve as the frontier moves.

For students raised on live feeds, static LibGuides feel like handouts, ones that look nothing like the Silicon Valley–coded websites and apps they use daily. A live-coded timeline feels more like a space they can inhabit and influence. It’s easy to imagine every course eventually having its own AI-native guide—a hybrid syllabus, resource hub, timeline, and AI studio—co-built by librarians, instructors, students, and agents. In this world, librarians don’t just maintain lists of databases. We design the environments where people learn to think and build with AI—not just consume static info.

Related Posts