What Is Vibecoding for Comms? The New Way PR Pros Build with AI

What Is Vibecoding — and Why Should Communications Pros Care?

In February 2025, Andrej Karpathy — one of the founders of OpenAI and former head of AI at Tesla — posted a simple observation: he was building software by describing what he wanted to an AI and letting it write the code. He called it "vibecoding."

The term stuck because it captured something real. For the first time in the history of software, you don't need to know how to code to build functional tools. You need to know what to build and how to describe it clearly. And that, my friends, is a communications skill.

I've been a PR professional for 23 years. I don't have a computer science degree. I couldn't write a Python function from scratch if you paid me. But in the last 12 months, I've built: a digital twin chatbot (the one on my homepage), an AI visibility audit tool, a GenAI press release builder, automated workflow tools, client dashboards, and a complete personal website — all by vibecoding.

This isn't about becoming a developer. It's about becoming a builder. And it's the single biggest unlock I've seen for communications professionals in two decades.

Why This Matters Specifically for Communications Professionals

Here's the thing most people don't realize about PR and comms work: we spend an enormous amount of time on tasks that are fundamentally pattern-based. Building media lists. Formatting press releases. Creating client reports. Monitoring coverage. Compiling analytics. These tasks follow predictable patterns — and patterns are exactly what AI can automate.

The traditional approach has been to wait for someone to build a SaaS tool that solves your specific problem, pay for it, and hope it fits your workflow. But here's the issue: nobody is building tools designed specifically for how communications professionals actually work. The tools we have were built by engineers guessing at our needs.

Vibecoding flips this. Instead of adapting your workflow to someone else's tool, you build tools that match your exact workflow. The person with the deepest understanding of communications work — you — becomes the person designing the solution.

This is why I've started calling myself a "Technology Builder" alongside "Communications Strategist." The two skills are converging. The most effective comms professionals in 2026 aren't the ones using the most tools. They're the ones building their own.

The Mindset Shift: From Tool User to Tool Builder

The mental shift is the hardest part, and it happens in three stages:

Stage 1: "I could never do that"

Most comms pros see code and immediately disengage. It looks foreign, complex, intimidating. This is normal. The key insight: you don't have to understand the code. You have to understand the problem.

Stage 2: "Wait, that actually worked?"

The first time you describe what you want in plain English and see working software appear in minutes, something shifts. You realize the barrier wasn't intelligence or education — it was access. Now you have access.

Stage 3: "I see tools everywhere"

This is where it gets exciting. Every repetitive task becomes a potential build. Every frustration with existing software becomes a project. You start seeing your entire professional life through the lens of "what could I build to make this better?" That's the vibecoding mindset.

"The best tool for your workflow is the one designed by someone who actually does your work. Now that someone can be you."

Real Examples: What I've Actually Built

Let me be specific about what vibecoding looks like in practice. These aren't hypotheticals — these are tools I've actually built and use daily:

GEO GPT — AI Visibility Audit Tool

What it does: takes a brand name and runs it through a structured audit across AI platforms, checking what ChatGPT, Gemini, and Perplexity say about the brand, identifying gaps, and generating a report with specific recommendations.

How I built it: described the audit process I do manually with clients to Claude, asked it to create a Custom GPT specification, iterated on the prompts until the output matched my methodology.

Time to build: about 4 hours across two sessions. The manual version of this audit takes 3-4 hours per client. Now it takes 5 minutes.

GenAI Wire Release Builder

What it does: takes a standard press release and restructures it for AI citation. Adds entity markup suggestions, restructures for extractability, includes FAQ sections, and formats for AI consumption.

How I built it: fed Claude our research on which press release formats get cited by AI (only 3 of 50 traditional releases we tested got cited), described the transformation rules, and iterated until the output matched our standards.

Command Center Hub

What it does: single-page dashboard that links to every doc, board, login, and tool a team uses. Searchable, categorized, always accessible.

How I built it: told Claude "I need a single HTML page that serves as a command center for a communications team. It should have sections for tools, docs, boards, and logins, be searchable, and look professional." First version was functional in 20 minutes.

DailyOps Email Digest

What it does: pulls activity from Slack, email, and Monday.com, synthesizes it into a single daily briefing email with action items highlighted.

How I built it: described the workflow to Claude, asked it to create a Node.js function that connects to each API, formats the digest, and sends it via email every morning. Deployed on Vercel.

Client Reporting Dashboard

What it does: automatically compiles media placements, coverage metrics, and AI visibility data into a visual dashboard clients can access anytime.

How I built it: described the data sources and desired visualizations to Claude, had it generate a React-based dashboard, connected the data sources, deployed it. Clients love it because it's always current — not a monthly PDF that's outdated the day they receive it.

Step-by-Step Getting Started Guide

Step 1: Pick Your First Project

Start small. Choose something you do manually at least once a week that annoys you. Media list compilation, report formatting, content template creation — anything with a repeatable pattern.

Step 2: Describe It in Detail

Write out — in plain English — exactly what the tool should do. Be specific. "I need a tool that takes a list of journalist names and beats, searches for their recent articles, finds their email addresses, and outputs a formatted spreadsheet" is much better than "I need a media list builder."

Step 3: Open Your AI Tool

Start with Claude (my recommendation for beginners) or ChatGPT. Paste your description and say "Help me build this." The AI will ask clarifying questions, propose an approach, and start generating code.

Step 4: Iterate Ruthlessly

The first output will be 60-70% right. That's normal. Test it, note what's wrong, and describe the corrections. "The output format is wrong — I need columns for Name, Email, Beat, Last Article, and Publication" is a perfectly valid iteration prompt.

Step 5: Deploy and Use

For web-based tools, deploy on Vercel (free tier is generous). For scripts, run them locally or set up a simple automation. The goal isn't perfection — it's functional and saving you time.

Tools and Platforms for Vibecoding

Here's what I recommend, ranked by accessibility for non-developers:

Claude (Anthropic)

Best for: Conversational building. Claude excels at understanding complex requirements described in natural language. It generates clean, well-commented code and explains what it's doing. My go-to for most projects.

Cursor

Best for: Multi-file projects. Cursor is an AI-powered code editor that lets you describe changes and navigate existing codebases. When your project grows beyond a single file, Cursor becomes essential. It understands context across your entire project.

Replit

Best for: Instant deployment. Browser-based coding environment with built-in AI. Write, test, and deploy without installing anything. Great for quick tools and prototypes.

ChatGPT with Code Interpreter

Best for: Data processing. If your project involves manipulating spreadsheets, analyzing data, or creating visualizations, ChatGPT's code interpreter can build and run the code in real-time.

Vercel

Best for: Hosting and deployment. Free hosting for static sites and serverless functions. When Claude builds you a web app, Vercel is where you put it so the world can use it.

What You Can Build in a Weekend

To demonstrate what's realistic, here are projects that genuinely take less than a weekend to vibecode:

• Personal brand website — Complete with AI chatbot, CMS-free, deployed on Vercel. (This entire site was vibecoded.)
• Media list generator — Input a topic/industry, output a formatted list of relevant journalists with contact info.
• Press release formatter — Convert standard press releases to AI-optimized GenAI Wire format.
• Client dashboard — Real-time metrics pulled from your existing tools, visualized in one place.
• Email template library — Searchable collection of your best-performing outreach templates.
• Meeting notes processor — Takes raw meeting transcripts and extracts action items, decisions, and follow-ups.
• Crisis communication plan generator — Input the scenario, output a structured response plan with stakeholder messaging.
• Social listening analyzer — Monitor specific topics across platforms and surface trending conversations.

Limitations and When to Hire a Developer

I'm bullish on vibecoding, but I'm not naive about its limits. Here's when you need professional help:

• Complex integrations: When you need to connect to APIs with complex authentication flows, enterprise security requirements, or real-time data streaming.
• Scale: If your tool will serve thousands of concurrent users, you need someone who understands infrastructure and performance optimization.
• Security-critical applications: Anything handling sensitive client data, financial information, or authentication needs professional security review.
• Long-term maintenance: If you're building something that needs to be maintained for years, having a developer establish proper architecture from the start saves pain later.

My rule of thumb: vibecode the prototype, then decide if it needs professional engineering. Often the vibecoded version is good enough. Sometimes it reveals that you need something more robust. Either way, you've saved time and money by clarifying your requirements before involving a developer.

The Future of Comms + Code

We're at an inflection point. The communications professionals who will thrive in the next decade are the ones who can both craft a message and build the system that delivers it. That's not a prediction — it's already happening.

I see three trajectories for the comms industry:

1. Comms professionals who ignore this — They'll remain valuable for their strategic thinking and relationships, but they'll be limited by their dependence on others to build their tools.
2. Comms professionals who vibecode — They'll have a significant advantage: faster workflows, custom tools, and the ability to prototype solutions in hours instead of months. This is the sweet spot for most people.
3. Comms professionals who become full builders — A smaller group who go deep into technology, building products and platforms that serve the entire industry. This is the path I'm on.

The beautiful thing about vibecoding is that it meets you where you are. You don't have to commit to becoming a developer. You just have to be willing to try describing what you need and seeing what happens. The AI handles the rest.

Sarah Evans

Communications Strategist & Technology Builder. 23+ years in PR, Partner at Zen Media. Everything on this site was vibecoded. Full bio →