What Happens When a Gold Mining Professional Tries Vibe Coding?

Source: Dev.to

The Problem I’m Trying to Solve

I work with assay reports—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 Issues

• Multiple labs, multiple formats – Labs in Ghana, Uganda, Tanzania, the UAE, and elsewhere send results in completely different ways.

  • Some 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.

• Inconsistent units, languages, and naming – Different units of measurement, languages (English, French, Arabic), regulatory standards, and sample‑naming conventions.

• Manual, error‑prone compilation – Every month my team spends days:

  1. Downloading reports from email.
  2. Opening each PDF individually.
  3. Copying data into Excel.
  4. Checking and converting units.
  5. Cross‑referencing sample IDs.
  6. Fixing inconsistencies.

By the time we’re done, the data is already outdated and we’re starting the next batch.

What I Actually Need

1. A system that can ingest assay reports regardless of format – PDF, Excel, email attachments, etc.
2. Standardisation – Convert everything to the same units, same structure, same fields.
3. Centralised, searchable storage – All data in one database that I can query.
4. Cross‑country and time‑period comparison – e.g., “Show results from Site A vs. Site B over the last six months” in minutes, not days.
5. Report generation – Quickly produce regulatory or investor‑ready reports on demand.

In short, I need to stop playing data‑entry clerk and actually analyse the information.

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. I’m sceptical, but also desperate enough to try.

• The problem is clearly defined and mostly about 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 £2,500+ per month.
• Enterprise mining software is overkill for our scale and would require months of implementation and training.

If AI coding tools can help me build something custom without learning JavaScript from scratch, that would be genuinely transformative—not just for me, but for anyone in a similar position.

My Extremely Rough Plan

1. Start with an AI coding assistant (Cursor or Claude Code) to scaffold a basic web app with a simple dashboard where I can upload files.
2. Data extraction –
   • Use OCR for PDFs.
   • Parse Excel files with varying column structures.
   • Handle multilingual text (English, French, Arabic).
3. Database – Evaluate Supabase or Firebase for storing the standardised results.
4. Visualization – Add charts to show grade distribution over time and comparisons between locations.

I honestly don’t know if this is realistic for someone with zero coding experience. That’s 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

• Immediate roadblocks – Will I hit a wall on the first technical hurdle and realise this whole thing was a fantasy?
• Security and data handling – Assay reports contain sensitive information about mineral deposits, site locations, and exploration results. Can I build something secure enough, or should I just pay for a proper solution?
• Time investment – I can’t spend six months learning to code. Could I have something working in a few weeks? A few months? What’s realistic?
• Robustness – Will AI tools fall apart when dealing with messy, real‑world data in multiple languages and formats, or can they actually handle the complexity?
• Maintenance and updates – If the system breaks, if lab formats change, or if we add new countries, do I need to understand the code to maintain it, or can AI tools continue to help?

Any advice, resources, or encouragement would be hugely appreciated. Thanks for reading!

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 business 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