---
title: "Internal tools: replace a spreadsheet with a real app in an afternoon"
description: "Turn a spreadsheet into an internal app with Lovable or Replit: the method, the prompt, access rules and real cases from Rokt and Productiv."
url: "https://vibe-code.blog/en/internal-tools-vibe-coding"
lang: "en"
author: "Thierry Gustin"
published: "2026-10-15"
updated: "2026-11-21"
---

# Internal tools: replace a spreadsheet with a real app in an afternoon

> **Key takeaways** — A vibe-coded internal tool is a good replacement for a shared spreadsheet when several people enter the same data, formula errors are expensive or each person needs a different view. In an afternoon you can import the CSV into a database, generate entry and tracking screens, and set who sees what. Rokt (135 apps in 24 hours on Replit) and Productiv (a dock-booking tool built in 10 days on Lovable) show what this looks like inside a company.

Internal tools are the most cost-effective place to start vibe coding at work: take a spreadsheet that three people fill in badly and turn it into a small app with forms, access rules and a dashboard. For a simple scope, one afternoon is enough to get a usable first version. Here's the method, the prompt and the limits you should know about.

## Why replace a spreadsheet with an app?

A spreadsheet becomes a problem when it's used as a shared database. The symptoms are always the same:

- people overwrite each other's entries;
- a broken formula skews a total and nobody notices;
- everyone sees everything, including the "salary" or "margin" columns;
- you need ten tabs and complicated filters so each person can find their view;
- the change history is impossible to follow.

An internal app fixes all five: a form controls input, the database holds one version of the truth, roles limit access, and each user gets their own screen.

## What companies are already doing

Two case studies published by the vendors show what's possible, with the caution this kind of source calls for.

- **Rokt** (e-commerce): according to Replit, more than 700 technical and non-technical employees presented 135 internal applications built in 24 hours during the company's hackathon. Examples include financial dashboards, a task engine that manages and prioritises more than 30,000 tasks a year, and legal document management tools. SVP Jon Humphrey explains the goal: "We set a goal to remove all friction that prevented us from operating 5x faster."
- **Productiv** (logistics, 1,200 employees across the US): according to Lovable, the team built ProDocks in 10 days, a dock-booking tool that replaced an off-the-shelf product costing $36,000 a year of which it used only a fraction. Another tool, ProPallets, replaced a paper process with barcode scanning and cut the team needed from 16 people to 4.

These cases illustrate the rule: the best internal tools aren't flashy. They fit a process the team knows by heart.

## Is your spreadsheet a good candidate?

Before you start, run your spreadsheet through this grid.

| Criterion | Good candidate | Keep the spreadsheet |
| --- | --- | --- |
| People entering data | 3 or more | Just 1 |
| How often it's used | Daily or weekly | Quarterly |
| Data structure | Uniform rows (one row = one customer, order, job) | Free-form calculations, what-if scenarios |
| Confidentiality | Some columns shouldn't be visible to everyone | Everyone can see everything |
| Entry errors | Frequent and costly | Rare |

If you mostly tick the left column, an app will save time. If your file is mainly a calculation and modelling tool, the spreadsheet is still the best tool in the world for that.

## The 5-step method for one afternoon

### 1. Clean and export the spreadsheet (30 minutes)

One header row, one row per record, no merged cells, dates in a single format. Export to CSV. If you use Supabase as your database, its documentation says you can import a CSV from the table editor, with a 100MB limit for dashboard imports.

### 2. Write the requirements sheet (20 minutes)

Who uses the tool, which roles exist, what each role sees and edits, and the 3 screens you can't do without. Our guide to [writing a spec your AI builder understands](https://vibe-code.blog/en/prd-for-ai-builders) gives the format.

### 3. Generate the app with a precise prompt (1 hour)

```text
I want to replace a Google Sheets spreadsheet with an internal app.
Here are the CSV columns: [list of columns with their types].
Users and roles:
- Admin: sees and edits everything, manages users.
- Team: creates and edits their own rows, sees their team's rows.
- Read-only: views the dashboard, edits nothing.
Screens:
1. A filterable, sortable list of records.
2. An entry form with validation (required fields, dropdowns
   for [fixed-value columns], enforced date format).
3. A dashboard: [3 specific metrics].
Constraints:
- Email login required, no public sign-up.
- Row Level Security enabled on every table according to the roles above.
- Audit log: who changed what and when.
- CSV export button for admins.
Propose the database schema first, then wait for my approval.
```

### 4. Import the data and test with a real colleague (1 hour)

Import the CSV, create one account per role, and ask a colleague to do their usual task without your help. Note every hesitation. For a testing method that doesn't require being a developer, see [testing AI-generated code](https://vibe-code.blog/en/testing-ai-generated-code).

### 5. Run both side by side for a week

Don't delete the spreadsheet on day one. A week in parallel uncovers the forgotten cases: the "notes" column everyone used for something else, the colleague who works on a tablet.

> **Start with the form** — If the afternoon isn't enough, ship the data-entry form connected to the database first. That's what eliminates errors, and it's the win the team feels immediately.

## Which tool for an internal tool?

| Tool | Strengths for internal tools | Watch out for |
| --- | --- | --- |
| [Lovable](https://lovable.dev/) | Database, login and roles generated by prompt | Invited collaborators use the owner's credits |
| [Replit](https://replit.com/) | Agent, database and deployment in one place, documented company use | Agent costs over time |
| [Bolt](https://bolt.new/) | Very fast prototype to validate the idea | Less suited to fine-grained permissions |
| Google AppSheet | Builds an app directly on Google Sheets, with Gemini | AppSheet licence needed beyond small teams |
| [Claude Code](https://www.anthropic.com/claude-code) or [Cursor](https://cursor.com/) | Code in your own repo, integrates with existing systems | Requires being able to read code |

Google describes AppSheet as a no-code platform whose Gemini assistant can build an app from a description, with Google Sheets as a data source. It's worth considering if your data already lives in Google Workspace and you don't want to move it. For a wider comparison, read our [guide to the best vibe coding tools](https://vibe-code.blog/en/best-vibe-coding-tools).

## The traps of vibe-coded internal tools

### Forgotten access rules

"Internal" doesn't mean "protected". An app published at a public URL with a misconfigured database exposes your data to anyone. Supabase's documentation says to enable RLS on every table in an exposed schema, and that once RLS is enabled, no data is accessible through the API with a publishable key until you create policies. Read our article on [the Supabase RLS mistake](https://vibe-code.blog/en/supabase-rls-security) before putting customer data in it.

### The ghost tool

The app built by someone who then leaves becomes a tool nobody knows how to change. Document who owns it, where the code lives (a company GitHub repository, not a personal account) and how to export it. Our guide to [Git for vibe coders](https://vibe-code.blog/en/git-for-vibe-coders) helps set that up.

### Personal data

If your spreadsheet holds employee or customer data, GDPR (in the UK and EU) applies to the app just as it did to the spreadsheet. Check where the tool hosts data and collect only what you need. We cover this in GDPR and your vibe-coded app.

### Scope creep

The first tool works, everyone wants their feature, and three months later the app is unmanageable. Keep a waiting list and add one feature at a time.

## Internal tools worth replacing first

- Leave and absence requests.
- A log of jobs or customer visits.
- An inventory of loaned equipment.
- Quote tracking and status.
- Booking schedules for rooms, vehicles or loading docks.
- An internal knowledge base with search.

You'll find more ideas in our [25 real vibe coding use cases](https://vibe-code.blog/en/vibe-coding-use-cases) and in our article on [automating your business with small homemade tools](https://vibe-code.blog/en/automation-vibe-coding).

## Frequently asked questions

## Frequently asked questions

### Can you really replace a spreadsheet with an app in an afternoon?

For a simple spreadsheet with a clean structure and three screens, yes: generation takes an hour, import and testing the rest. What takes longer is the week of parallel use to find forgotten cases. A spreadsheet with dozens of tabs and nested formulas will need several sessions.

### Do you need to code to build an internal tool?

Not for a first version with Lovable, Replit or AppSheet. You do need to understand roles and access rules, and how to check that a "read-only" account truly can't edit anything. If the tool touches sensitive data, have someone technical review the database configuration.

### Does my data stay in Google Sheets?

It depends on the tool. AppSheet can work directly on Google Sheets. Lovable and Replit usually create a real database and you import your CSV into it, so the spreadsheet becomes an archive. A dedicated database is more reliable for simultaneous access and permissions, but you have to accept a change of habit.

### How much does a vibe-coded internal tool cost?

The main cost is the vibe coding tool subscription and the credits used during iterations. Lovable's documentation states that invited collaborators use credits from the project owner's workspace. Compare that cost with the subscription for the off-the-shelf software you're replacing, as Productiv did.

### Does an internal tool need the same security as a public app?

Yes, if it's reachable from the internet. An internal app at a public URL is exposed exactly like a consumer app. Mandatory login, no open sign-up, RLS enabled on every table and no secret keys in the front end: the same rules as in our security checklist.

## Sources

- [How Rokt built 135 internal applications in 24 hours with Replit — Replit](https://replit.com/customers/rokt)
- [Productiv customer story — Lovable](https://lovable.dev/customers/productiv)
- [Import data into Supabase — Supabase Docs](https://supabase.com/docs/guides/database/import-data)
- [Row Level Security — Supabase Docs](https://supabase.com/docs/guides/database/postgres/row-level-security)
- [AppSheet — Google](https://about.appsheet.com/home/)
- [Collaboration — Lovable Documentation](https://docs.lovable.dev/features/collaboration)

— Thierry Gustin, Formateur en intelligence artificielle (GustinAI) · November 21, 2026
