In the first quarter of 2026, AI wasn’t a side topic – it was part of the regular workday at Levi9. From workshop sessions and informal discussions, to published technical articles and an internal challenge, AI made its way into projects, processes, and the everyday conversation within the community.
TuesdAI: Workshops That Keep Going
TuesdAI – a workshop series held every Tuesday at Belgrade and Zrenjanin – continued to grow this year. The initiative is driven by the Competence Development team, and sessions cover concrete tools and topics that can be applied to work immediately.
In January and February, colleagues shared knowledge directly relevant to their daily work: Aleksa Stanić kicked off the year with a session on the Firebender tool in Android Studio, Stefan Milankov presented how knowledge leads and AI accelerates through a session on building a data system, Igor Janković covered GitHub Copilot tips, tricks, and MCP in practice. In March, Dušanka Lečić spoke about observability as a superpower in the age of AI, and Ivan Vuković showed us how, using MCP tooling, an AI model can be given creative freedom to act as an interior designer- while still adhering to the laws of physics.
The format is straightforward and effective: every week, a new speaker, a new topic. Participant feedback shapes the sessions that follow – topics come from the people who use these tools and want to understand them better.
AI Hub: From Talks to Conversations
Running alongside TuesdAI, the AI Hub at the Novi Sad office serves as a space for exchanging experience and knowledge around the practical use of AI.
In January, Marko Šuker, Data Architect, held a hands-on session dedicated to the RAG approach – with an ELI5 breakdown of key LLM engineering concepts and a practical implementation walkthrough. The article that came out of that session, published as part of the AI9 series, goes a step further: it makes the case for why RAG, despite claims that it’s “dead,” remains one of the most relevant architectural patterns in production AI systems.
In March, an informal discussion was held on the topic of AI in the software development lifecycle (SDLC). The discussion was facilitated by Ružica Kresoja, and a large number of colleagues gathered in the office to exchange thoughts and experiences. Kristina Savić shared how Levi9 plans to use metrics from Cursor and GitHub Copilot to better understand the impact of AI tools on team productivity.
The AI9 Series: Knowledge That Travels Further
Alongside internal activities, Levi9 published a series of articles as part of the AI9 series – an initiative where colleagues share concrete experiences from client projects and workshop sessions.
Nebojša Ložnjaković and Goran Militarov wrote about the Model Context Protocol (MCP) – an open standard for connecting AI applications to external systems – sharing insights from a real-world implementation for a client whose goal is to become “AI first.” Igor Janković, through his piece on AI orchestration in software development, described the shift from writing code to managing AI agents and validating their output. Igor Dedić presented a talent discovery application built on Computer Vision and a proprietary 3D Pose Engine, designed to make scouting accessible to 300 million young footballers around the world. A team of authors – Dragan Škrinjar, Nebojša Tamindžić, Ljubica Turanjanin, and Željko Miladinović – described an intelligent incident detection system for multi-cloud environments that doesn’t just alert, but analyzes, prioritizes, and routes the response.
AI Challenge: Innovation Without Role Boundaries
In February, Levi9 ran an internal AI Challenge – a two-week initiative in which 22 colleagues developed complete AI solutions, from concept to video pitch. What set this format apart was the diversity of participants: alongside developers, the challenge included DevOps engineers, QA specialists, and Delivery Managers. Each person identified a real business problem and built a practical AI-driven solution.
The winners were Slađana Tufegdžić and Jovan Zejnula – but the real outcome of the Challenge was something broader: proof that with the right framework and support, innovation can come from any role.
Building the Habit, Not Just the Hype
What Q1 2026 showed is that AI adoption at Levi9 isn’t driven by obligation or trend-chasing – it’s driven by curiosity, peer learning, and a genuine interest in building better. Workshops, discussions, published insights, and internal challenges are all part of the same effort: making AI a natural part of how we think and work. The tools will keep changing. The habit of learning together is what stays.
In this article:
