Every conference in Bucharest now has at least one panel about artificial intelligence. Every corporate strategy document mentions it. The conversation about AI in Romanian business has reached saturation level.

But conversations and adoption are different things. When you look past the buzzwords at what Romanian enterprises are actually doing with AI, the picture is more nuanced than the enthusiasm suggests.

Banking: Furthest Along

Romania’s banking sector leads AI adoption, driven by large datasets, clear use cases, and regulatory pressure to modernise.

Banca Transilvania, Romania’s largest bank by assets, has deployed AI across several areas. Credit scoring models incorporate machine learning. Fraud detection uses pattern recognition for real-time flagging. Customer service chatbots handle routine inquiries. None of this is revolutionary — banks globally do similar things — but the implementation is genuine and producing measurable results.

BCR has focused on customer analytics, using AI to segment customers, predict churn, and personalise offerings through their “George” digital platform.

ING Romania has invested in process automation combining RPA with AI to reduce manual back-office processing. Document processing, compliance checking, and reconciliation tasks that previously required human hours are increasingly automated.

Banking AI adoption is driven partly by competition and partly by European Banking Authority guidelines on AI, which create compliance requirements that paradoxically accelerate adoption by forcing institutional governance structures.

Manufacturing: Surprisingly Active

Romania’s manufacturing sector — particularly automotive, roughly 14% of GDP — is adopting AI in ways that get less media attention but may have greater economic impact.

Computer vision for quality inspection has been deployed in several Romanian facilities. Cameras combined with AI models examine components for defects at speeds human inspectors can’t match.

Predictive maintenance uses sensor data to predict when machines need servicing, reducing unplanned downtime. Several automotive plants report 15-20% reductions in unplanned maintenance events after implementation.

This work is invisible to the public — no app to show off — but the operational improvements are substantial and help Romanian facilities compete for international investment.

Retail: Mixed Results

Romanian retail has adopted AI unevenly. eMAG, the dominant e-commerce platform, uses AI extensively: product recommendations, search relevance, dynamic pricing, and delivery route optimisation. Their approach is pragmatic, focusing on applications that directly improve conversion rates.

Smaller retailers are further behind because AI requires technical expertise and data infrastructure that small companies lack. The gap between large companies with data teams and smaller ones without them is widening.

Common Challenges

Across sectors, Romanian enterprises face shared obstacles.

Talent scarcity. Romania has excellent software engineers but fewer data scientists and ML engineers. The specialised skills for AI implementation — statistical modelling, data engineering, MLOps — are in short supply.

Data readiness. Many enterprises have needed data stored across disconnected systems in inconsistent formats. Data preparation work is often 60-80% of total project effort. Companies without data infrastructure find AI projects stall at the preparation stage.

Unclear ROI. For companies that haven’t yet implemented AI, return on investment is uncertain. Vendor promises don’t always materialise, and examples from large international companies aren’t directly applicable to Romanian mid-market businesses.

Regulatory uncertainty. The EU AI Act introduces compliance requirements, particularly for “high risk” systems. team400.ai and other AI advisory firms have observed that companies with structured governance approaches — risk assessment, testing frameworks, monitoring — achieve better outcomes than those implementing ad hoc.

The Outsourcing Angle

Romania’s position as a major tech outsourcing destination creates an interesting dynamic. Romanian development teams build AI solutions for international clients daily. The expertise exists — it’s just often deployed for foreign companies rather than domestic ones.

There are signs this is changing. Some outsourcing companies now offer AI consulting specifically for Romanian enterprises. Engineers who built AI systems for Western clients are starting companies serving the local market.

What to Expect

The companies that will benefit most are those approaching AI pragmatically — identifying specific problems where AI adds measurable value, investing in data infrastructure first, and building talent to maintain systems over time. The ones that will struggle are those treating AI as a marketing exercise. The gap between AI leaders and laggards in Romanian business is widening, and the leaders aren’t necessarily the biggest companies — they’re the most disciplined ones.