The AgentForge hackathon, organized by SoftServe's technical teams — the Gen AI Lab team and the Data Science department — recently concluded. After an intensive online stage, an offline finale was held in the company's Lviv office. The hackathon brought together developers, ML engineers, Data Science specialists, and researchers working as autonomous agents.
The AgentForge hackathon, organized by SoftServe's technical teams — the Gen AI Lab team and the Data Science department — recently concluded. After an intensive online stage, an offline finale was held in the company's Lviv office. The hackathon brought together developers, ML engineers, Data Science specialists, and researchers working as autonomous agents.
35 teams and 6 solo participants competed for victory in the hackathon, developing AI agents over several days, following several requirements — from the technical features of the solutions to the possibilities of their practical application.
"The AgentForge hackathon became a good platform for working with autonomous systems that are capable of not just responding to requests, but acting independently. We saw many interesting ideas and strong teams that were not afraid to take on real complex technical cases. The participants managed to organize the work of the teams in a short time and present their prototypes to the community and the jury," says Nazar Drushchak, AI Technical Consultant at SoftServe.
The winner of the hackathon was the CognitiveOps++ team, which received a prize of $5,000, presenting its project ADW (Agentic Data Workflow) — a production-ready AI platform that transforms business queries into certified analytics without the involvement of large BI teams. The goal of the project is to reduce time-to-insight from weeks to minutes. ADW combines AI, data engineering, and analytics into a single, managed platform for real business impact. The solution helps businesses gain insights from data faster, and data teams move from operational support to a strategic role.
According to the jury, the project was distinguished by its practical value, mature architecture, and clear demonstration of how autonomous AI agents can automate complex data processes in real-world business scenarios.
Second place and $3,000 went to the No More Tokens team with the AI Product Manager Colleague project. The AI agent is able to analyze “raw” documents (PDFs, specifications, presentations), understand the existing code base, and provide reasoned answers. It doesn’t just generate text, but helps assess technical complexity, find common ground in the code, and formulate clear product requirements. The AI agent helps write product requirements and checks whether the code meets them. The team tried to automate a complex and chaotic part of the process — coordination between technical and non-technical teams.
Shalena Bdzhilka’s team took third place and won $2,000 with their project Machine Learning & Data Analytics Agent. The team created a tool for structuring ML experiments. The system uses multiple specialized agents that work together to explain results, maintain order in processes, and help users solve ML problems.
