RUBICON Software Development

15/10/2026

Is your planning file called "plan_final_v3_updated"? We need to talk. 🫠

Excel works well for one person and one model. Trouble starts when sales, finance and operations each keep their own files, and only one colleague knows how they connect.

In the article:
🚩 7 signs your process has outgrown spreadsheets
🧩 How "SE" and "Southeast" can break a brand new system
🧹 The four cleanup steps to finish before anyone builds a screen

📖 Read it here: https://www.rubicon-world.com/blog/signs-your-process-has-outgrown-excel

Signs your process has outgrown Excel, and why cleaning and unifying data first turns spreadsheet sprawl into one source of truth and live simulations.

29/09/2026

Your dashboard says revenue is up. How do you know it's right? 🤔

For most data teams, the honest answer is "we eyeball it and compare it to last week."

Haris Tiro, QA engineer at RUBICON, explains how to test data the way software teams test code:

→ Five checks that catch bad numbers before anyone sees them
→ The silent unit change that made one field 1,000× too big (and the check that caught it)
→ Why fixing this early is cheap, and fixing it late is an audit

Read the full article 👉

Software has test suites. Data has stakeholders who spot bad numbers too late. See how automated data quality checks catch silent errors before they ship.

24/09/2026

Is your data actually ready for AI? 🤔

Most teams find out the expensive way: the model works fine, the answers don't.

Dženana broke it down into four levels, from "the answer lives in someone's inbox" to "every answer cites its source":

→ How to tell which level you're on
→ The quality check almost everyone forgets (freshness)
→ How one leadership team went from hours of meeting prep to seconds

Read the full guide 👉 https://www.rubicon-world.com/blog/is-your-data-ready-for-ai-and-how-would-you-know

22/09/2026

"Buy, don't build" is usually the smart call. This was the exception.

A large enterprise was paying for an off-the-shelf digital twin to map its supply chain, but with thousands of products it was slow, clunky, and expensive. We built a custom platform instead: a graph database on Azure that renders the whole network and lets anyone trace a material through the product line.

Five months to deliver. Owned outright. Costs less than the license it replaced.

Read the full story: https://www.rubicon-world.com/cases/supply-chain-insights-through-digital-twin-analytics

17/09/2026

Where does the money go on an analytics project before a single chart shows up? 📊

Faris Karčić went digging through one real case, a manufacturing company drowning in years of Excel:

→ Same report, three years, three different spreadsheet layouts
→ Rows that Excel happily accepted and the pipeline flatly refused
→ A "static" data table that turned out to be anything but

Spreadsheets let you do whatever you want. Pipelines do not. That gap is where most of the budget lives.

Read the full story 👉 https://www.rubicon-world.com/blog/analytics-project-budget-before-first-dashboard

The Hidden Cost of Legacy Data in Data Engineering Projects

15/09/2026

Our engineers built an AI assistant for a leadership team drowning in meeting notes and status updates. The interesting part is what they refused to do.

The standard approach lets an AI read your documents and build the knowledge graph on its own. The team tried it. It made up job roles and turned one real person into three separate records. Not something you want a CEO making decisions on.

So they built the foundation by hand instead. People, roles and projects are verified by humans, and the AI only reasons over that verified layer. In the proof of concept, prep time dropped from hours to seconds.

Swipe through the slides for the full architecture, including the tradeoffs (slide 8).

Full case study: https://www.rubicon-world.com/cases/graphrag-neo4j-enterprise-chatbot

Where does your team's knowledge live right now: Slack, a wiki, or in someone's head? 👇

25/08/2026

Microsoft doesn't hand out partner badges for showing up.

RUBICON is now a Microsoft Solutions Partner for Cloud & AI Platforms, earned through certified staff, delivered Azure projects, and actual customer usage on the platform.

Read the full story 👉

RUBICON is now a Microsoft Solutions Partner for Cloud & AI Platforms, proof of delivered Azure cloud and AI work across banking, healthcare, and manufacturing.

19/08/2026

In 1997, Woody Harrelson played an American journalist trying to get the world to see what was happening to Sarajevo during the siege.

Last night, almost thirty years later, he was back. Sarajevo Film Festival screened Welcome to Sarajevo at the open-air cinema, and Harrelson came in person to receive the Heart of Sarajevo, the festival's top award.

Our team was there for it, a few seats away from the man on screen.

We come back to this festival every year because Sarajevo is home, and nights like this remind us why.

16/07/2026

We're back from GITEX AI Europe 2026, where RUBICON once again joined the Bosnia and Herzegovina Pavilion alongside some of the country's leading tech companies.
Great conversations, great energy, and a lot of pride in showcasing what BiH's tech sector can bring to the global stage.
Thank you to everyone who stopped by to connect with us.

03/06/2026

Plot twist: your code has a carbon footprint. 🌱

A few things worth knowing if you're in a tech or engineering role:
→ Idle cloud VMs burn up to 60% of peak energy doing nothing
→ Training GPT-3 emitted about 500 tons of CO₂e
→ A 30% CPU efficiency gain saves tens of tons of CO₂e a year

The good news: greener code is just faster, cheaper, leaner code.

Esmina wrote the no-buzzword guide for engineers 👉

A practical guide to sustainable software engineering: life cycle assessment, CSRD, the GHG Protocol, and the green coding practices that cut the carbon footprint of your code and cloud.