Newstodon

Newstodon provides news and trends about businesses in Romania. We offer both balanced and sensationalist perspectives on what’s happening in the Romanian business world—new company formations, closures, dissolutions, and other key trends.

Why does this site exist? Because journalism about company statistics is typically shallow and sensationalist. I set out to demonstrate that AI could deliver much better analysis—and also show how it could be dramatically worse if misused. The dual narrative approach (balanced and sensationalist) exposes both the potential and the pitfalls of LLM-generated content.

Data Sources

We track:

• New business registrations - companies starting operations
• Business closures and changes - suspensions, dissolutions, and deregistrations showing when companies wind down
• Data comes from the National Trade Register Office (ONRC), Romania’s official business registry

Methodology

Our analyses are based on:

• Monthly and year-over-year (YoY) comparisons
• Analysis by economic sectors (CAEN classification)
• Geographic analysis by county and region
• Legal entity types (SRL, PFA, CA, etc.)

Generated Articles

Newstodon generates articles with two narrative tones:

• Balanced (News) - Factual analysis with context
• Sensationalist (Crisis) - Alarming and dramatic perspectives

Technology

The platform uses:

• Excel data parsing from official registries
• SQLite database for storage and analysis
• LLM agent (DeepSeek) with tools for statistical computation and contextual article generation
• Hugo for static site generation