What Happens When a Gold Mining Professional Tries Vibe Coding?

Hi All

I'm not a developer. I work in the gold sector across multiple African countries and the UAE, and I spend an absurd amount of time wrestling with spreadsheets and PDFs that should be talking to each other but aren't.

Everyone keeps saying AI can code now. That tools like Cursor and Claude can help non-developers build actual applications. I've been sceptical, but I'm at the point where I need to either test this myself or keep wasting days every month on manual data compilation.

So here's what I'm going to try, and I'm documenting it from the start because I genuinely don't know if this will work or if I'm being ridiculously naive.

The Problem I'm Trying to Solve

I work with assay reports. These are laboratory analyses that show mineral content in rock and soil samples. These reports are critical for determining whether a site is worth developing, what the ore grade is, and where to focus our efforts.

The issue? I'm dealing with labs across Ghana, Uganda, Tanzania, the UAE, and several other regions. Each one sends results in completely different formats.

Some labs send PDFs. Others send Excel spreadsheets with varying column structures. Some provide proprietary lab software exports. One lab still occasionally faxes reports that need manual entry.

On top of the format chaos, there are different units of measurement, different languages (English, French, Arabic), different regulatory standards, and different naming conventions for samples.

Right now, my team spends days every month manually compiling these into one master spreadsheet so we can actually compare results across sites. It's error-prone, tedious, and an absolute productivity killer.

Every month, the same routine. Download reports from emails. Open each PDF individually. Copy data into Excel. Check units. Convert measurements. Cross-reference sample IDs. Fix inconsistencies. By the time we're done, the data is already outdated and we're starting the next batch.

What I Actually Need

I need a system that can ingest assay reports regardless of format. PDF, Excel, email attachments, whatever the lab sends. It needs to standardise the data into the same units, same structure, same fields. Everything stored in one searchable database where I can actually find what I need.

The system needs to allow comparison across countries and time periods. If I want to see how results from Site A compare to Site B over the last six months, I should be able to pull that up in minutes, not spend two days building a comparison spreadsheet.

It needs to generate reports when we need them for regulatory compliance or investor updates. Right now, every request for data turns into a multi-day project.

Basically, I need to stop playing data entry clerk and actually analyse the information. That's what I was hired to do, but I spend most of my time doing manual data work that software should handle.

*Why I Think AI Coding Tools Might Help
*
I've been hearing about "vibe coding". The idea that you can describe what you want and AI tools help you build it even if you don't know how to code. Sceptical? Absolutely. But also desperate enough to try.

Here's my thinking. The problem is clearly defined. It's mostly data transformation, which seems like something software should handle. I can articulate what I need, even if I can't write a single line of code.

Off-the-shelf solutions either don't exist for this specific use case or cost upwards of £2,500 per month. Some enterprise mining software exists, but it's designed for massive operations and would require months of implementation and training. For our scale, its complete overkill.

If AI coding tools can actually help me build something custom without learning JavaScript from scratch, that would be genuinely transformative. Not just for me, but for everyone else in similar positions who have clear problems but no technical skills.

My Extremely Rough Plan

From what I've researched, mostly by reading posts here on dev.to, I'm thinking I should start with Cursor or Claude Code to help me build the basic structure. Create a simple dashboard where I can upload files.

Then figure out data extraction. This is probably the hardest part since formats are all over the place. I've read about OCR for PDFs and parsing Excel files, but I have no idea how complex this actually is.

Build a database to store standardised results. I've heard of things like Supabase and Firebase. Are those appropiate for this? I don't know.

Add basic visualisation so I can actually see trends across sites. Maybe some charts showing grade distribution over time or comparison between locations.

I honestly don't know if this is realistic for someone with zero coding experience. That's part of what I'm trying to find out. Can these AI tools actually bridge the gap between "I know what I need" and "I can build it"?

What I'm Worried About

Will I hit a wall immediately? Everyone says these AI tools are amazing, but will they actually help when I have zero programming knowledge? Or will I get stuck on the first technical hurdle and realise this whole thing was a fantasy?

Security and data handling. These assay reports contain sensitive information about mineral deposits, site locations, and exploration results. Can I even build something secure enough to handle this data properly? Or should I just pay for a proper solution and accept the cost?

Time investment. How long does something like this actually take? I can't spend six months learning to code. But could I have something working in a few weeks? A few months? What's realistic?

What if it just doesn't work? Will AI tools fall apart when dealing with messy, real-world data in multiple languages and formats? Or can they actually handle the complexity?

Maintainance and updates. Even if I build something, what happens when it breaks? When lab formats change? When we add new countries? Do I need to understand the code to maintain it, or can AI tools help with that too?

Why I Joined This Community

I'm planning to document this journey, the successes and the inevitable failures. If AI coding tools actually live up to the hype, I want to show what's possible for non-technical people solving real buisness problems.

And if it doesn't work? Well, that's useful information too.

So, any advice before I get started on this?

Thanks a lot for reading!
Marcus Briggs

This is about all I can do so far with AI is get images ...lol