Guide to Vibe Coding for People&Org

Oct 10

Preface: A Weekend and a Working App

One quiet Saturday afternoon, I decided to test myself: could I, an HR Executive, who hadn’t written code since the early 2000s, still build something real?

I called a friend and explored vibe coding. It’s a way to build tools using AI as shadow software developer.

Four hours later, I had a tic tac toe web app. But the app wasn’t the point, the workflow was the most important lesson to me.

I realized this same approach could transform how we do HR analysis, modeling, and dashboards, or essentially everything we currently juggle in Excel.

Now, I must say I had a strong motivation. I work for a pre-commercial life science company and efficiency and lean operations is a name of the game. When I can’t add people to my team, I have to rely on digital tools and ideas to essentially perform tasks I need.

What Is Vibe Coding?

Vibe coding is essentially a collaborative creation loop where you act as a manager of few very capable but inexperienced coders, all of them are AI. If you are familiar with behavioral therapy, it’s a bit similar: if you want to teach a child some complex skill, you have to break it down to simpler skills buildout. Also, AI hallucinates, and it’s helpful to have different LLMs review each other results.

Describe what you want.
Let the AI draft a plan.
Have another AI review it.
Iterate for clarity before implementation.

You don’t need to be an engineer, just be a structured inquisitor. If you ever had to create an SOP, it’s the same thinking style. You can use vibe coding to transform “spreadsheet logic” into systems: dashboards, compensation modeling tools, headcount visualizations, and even automated workflows.

Why People& Org or HR Should Care

We already have to think like builders: we model scenarios, analyze data, and project outcomes. But we do it in Excel - cozy and familiar, but static, manual and high-maintenance when it comes to version control.

Vibe coding may allow HR teams:

Replace manual reporting with automated dashboards (if your master data source allows it).
Build lightweight apps for compensation modeling, headcount forecasting, analyzing benefits, or survey results.
Develop interactive analytics without waiting for IT or vendors.
Use AI as a partner in design, not just as a chatbot. Well, partner might be a strong word… Junior assistant?

It’s a new way to bring analytical thinking to life: make operational HR strategy through intelligent prototypes.

The Vibe Coding Workflow (Simplified for HR Projects)

Describe what you need with specifics.
Example: A dashboard that tracks turnover by department and flags attrition risks by pay position at the market (those who are below 25th percentile) and have higher tenure at the same position (3+ years at the same position).
Ask an AI to draft a PRD (Product Requirements Document).
This becomes a blueprint, it contains goals, data inputs, outputs, and success metrics. Now, AI is the most skilled at software engineering and development, so, in my experience, it performs at it’s best when you use engineering language.
Have another AI review it.
Use a different model for quality control, ask for missing scenarios or edge cases. Even if you don’t know what edge case is, just ask the question.
Ask your first AI to Integrate that feedback.
Update until every HR analyst could implement from it without guessing. It’s also important to ask any of AIs to check for over-engineered areas. Imagine AI as a fresh grad with best grades and zero knowledge of how the real world works, they try to incorporate all academic knowledge they have into every simple task and answer questions you didn’t ask to impress you. Like if you ask it to analyze compensation equality across the data set, it will try to bring the country issues with salary equality across everything if you don’t ask it specifically not to do it, and focus on dataset.
Translate the plan into system architecture.
Ask AI to translate your PRD into technical architecture. Ask to keep it readable, name components like “data import,” “logic,” and “UI charts.”
Get a second review for simplicity and risk.
Again ask to QC for data exposure, over-engineering, or excessive dependencies.
Turn plans into a to‑do list.
Ask AI to create a To-Do list based off your technical architecture. It should create a list of specific and relatively simple tasks grouped by themes.
Start building through conversational coding tools.
Feed all three documents you have (PRD, Technical Architecture, and To-Do list) to SW development AI like Manus, Claude Code, Cursor or ChatGPT for implementation, step by step. You might need paid model to make it work better. But free version, or open source one may work if your task isn’t complex. Tools like Manus may offer you to develop a web page, which is useful for QC, when you can test the outcome and provide feedback immediately. In my experience, if you have design preferences in mind (e.g. background, pictures and etc), put them in from the start. Most tools struggle to implement design changes afterwards.
Review and test.
Again alternate AI tools for peer review, track reasoning, and run functionality checks.
Add automation testing and human simulations.
Test both “normal” and “failure” HR use cases (e.g., missing data or updated org charts).

Deployment: Bring It into the Real World

Once your app or dashboard works and passes testing, deployment doesn’t need to be complex. For most HR applications, a corporate hosting environment such as SharePoint or an internal web server is perfectly sufficient.

Before publishing, always check with your IT department to confirm:

The hosting environment meets company security standards.
Access is limited to the right roles or groups (especially if data is sensitive).
Encryption and authentication match company policy.
Data sources are approved and properly backed up.

Even simple dashboards can contain compensation or organizational data, so a quick IT security check early on can prevent major risks later.

Once cleared, your vibe coded HR apps can live safely beside your other corporate tools, accessible through familiar links or portals.

Best Practices for HR Builders

Don’t code until the PRD passes two review loops.
Read PRD! Give to your colleagues to read! Read it again! AI communicates using natural language, so your brain will make you think it understands you and you can trust it. It does not understand you and you can’t trust it.
Keep every system box explainable in one sentence.
Break large tasks into day sized actions.
Define “done” as reviewed, tested, and reliable for others.

What Good Feels Like

When vibe coding clicks, it feels rhythmic. You draft, refine, test, and deploy with confidence.

You’re not just maintaining spreadsheets, you’re building HR systems that adapt and grow. The process turns creativity and analysis into continuous improvement.

For me, it was proof that you don’t have to be a developer to build something powerful. You just need structure, investigative mindset, and a willingness to learn

Alena Reva