AI in Poland

Decathlon replaced an inflexible central customer service system with Zowie's AI Agent and AI Inbox, combining chat, email and an integrated hotline in one platform. Deflection rose from 30% to 50%, customer service efficiency improved 16%, CSAT reached 4.6, the AI covered the work of 19 extra seasonal agents, and support-driven revenue grew 20%.

ALGORCOMP built a custom AI order pipeline (OpenAI GPT-4o + Node.js + React) for Polish stoneworker GRUPA KRES — consolidated 6 disconnected tools into one system handling 100 monthly orders; time from inquiry to quote dropped from 5–7 days to 1 day, margin per order +6 pp within 6 months, customer NPS +28 points in Q1.

Generali Poland deployed "Leon," an AI virtual assistant built with IBM watsonx Assistant, to handle 120+ common customer support scenarios 24/7. The chatbot now handles more than 97% of customer conversations without human redirection, saving 120 consultant hours per month and managing 6,500 conversations monthly.

UNIQA Poland deployed an AI Knowledge Assistant built on the Polish language model PLLuM and a strict RAG pipeline to give agents real-time, verified answers drawn from thousands of insurance documents. The on-premise system integrates with CRM and Call Center, runs across the company's nationwide agent network, and was delivered by RITS in roughly eight weeks.

Poland's National Revenue Administration (KAS, Krajowa Administracja Skarbowa) under the Ministry of Finance deployed an AI-based toll enforcement system in 2021-2022 across 40 toll stations and 102 patrol cars on the A2 and A4 motorways, achieving over 98% automatic number plate recognition reading rates and almost 100% accuracy for human-readable plates. The deployment replaced Poland's old stop-and-go toll collection with a free-flow satellite-based e-TOLL system.

Gdańsk shipyard Remontowa uses Ship Spec Reader, an NLP and machine-learning application from MLNative that pulls material requirements, task descriptions, dimensions and compliance details out of unstructured ship repair specifications. The vendor reports an 80% cut in data processing time, with no published baseline.

Kingfisher's Global Business Services hub in Kraków uses Rossum to read 40,000 invoices a month for six countries and pass the data to RPA bots that index it in SAP. Since going live in January 2021, 79% of fields are read automatically, 60% of invoices are touchless, and 14 FTEs were redeployed.