In August, teams from around the world supercharged their agentic AI skills and collaborated on innovative solutions as part of the second annual IBM TechXchange Pre-conference watsonx Hackathon. This exciting, 17-day virtual event brought the developer community together to dive into the power of AI agents and build cutting-edge solutions that tackle real-world problems.
Teams were challenged to build proof-of-concept solutions using IBM watsonx to show how agentic AI can take on an everyday issue in a new, creative way. They could pick between two hackathon themes:
The top 2 hackathon teams each won $5,000 USD and a full trip to the IBM TechXchange 2025 conference in Orlando, Florida in October!
Winning teams
First place for the "Build with agentic AI challenge":
Team AIght
In fields like finance, healthcare, legal compliance, and customer support where critical decisions so often depend on the accuracy of verbal communications, important claims made during meetings or calls often go unverified. For example, a customer service representative claims a service has “99.99% uptime.” These statements, if untrue or unsupported, can result in not just reputation damage, but legal liabilities, regulatory penalties, or financial loss. Existing meeting transcription and summarization tools focus on creating readable notes, but do not actively cross-check the validity of claims against authoritative knowledge bases.
The "ClaimCheck AI" solution is an agentic AI system that automatically transcribes recorded or live meeting conversations with speech-to-text, and then extracts possible factual claims made within the resulting transcript, retrieves relevant evidence from a curated knowledge base using IBM watsonx embeddings and re-ranking, verifies whether each claim is supported, contradicted, or unverifiable, and finally summarizes the findings in a structured, auditable report. This approach moves beyond passive transcription to active truth-verification, in real time.
The solution uses a pipeline of multiple, IBM watsonx.ai-powered agents, with dedicated agents for transcription, claim extraction, retrieval, verification, and summarization. And the knowledge base is scalable, so that it can grow to support additional industry-specific compliance rules, SLAs, and historical records.
First place for the "Sustainable cities and communities challenge, brought to you by Call for Code":
Team Hive.ai
Cities face an urgent climate adaptation crisis but lack the tools to act effectively. Municipal departments struggle with fragmented climate data, slow manual analysis, and disconnected sustainability frameworks. A flood risk assessment might take months to complete, while heat wave projections sit isolated from infrastructure planning. Meanwhile, thousands of successful climate interventions exist globally, but cities can't efficiently identify which solutions fit their specific context, budget, and UN Sustainable Development Goal (SDG) commitments. The result? Cities react to climate disasters instead of preventing them. Resources are misallocated. Proven solutions remain undiscovered. International funding opportunities are missed because projects lack proper SDG alignment documentation.
The "Climate AI Hive" solution organizes climate change data from multiple sources into actionable urban strategy and democratizes access to climate expertise. A user can input any city name, and specialized BeeAI agents retrieve historical and projected climate data from multiple scientific models and official UN SDG data for the target city and region, assess trends, patterns, and vulnerabilities, and formulate actionable, locally adapted recommendations in mobility, waste, green spaces, energy, and citizen engagement. Proposed interventions are evaluated against the UN SDG 11 framework, identifying relevant targets and indicators, scoring their alignment, and suggesting improvements to maximize compliance and funding potential. In just minutes, the platform produces an integrated resilience report with climate outlooks, prioritized actions, implementation guidance, budgets, timelines, and SDG 11 alignment metrics.
Through deployment on IBM Cloud, the team hopes to continue developing Climate AI Hive and present it at the United Nations ITU AI for Good Conference 2026.
Congratulations as well to our second place winners. Each member of the second place teams will take home Apple Airpod Pros® and $300 USD IBM Cloud Credits, as well as complimentary IBM TechXchange 2025 conference passes. In second place:
-
Build with agentic AI challenge: Team Shanti — This agentic AI-powered career dashboard provides a holistic way to approach job hunting and career advancement
-
Sustainable cities and communities challenge, brought to you by Call for Code: Team TechGirl — This agentic AI solution creates walking maps rated by safety to help cities provide safe, accessible transport for all
This could be you!
IBM TechXchange is hosting a series of worldwide, virtual hackathons throughout 2025! Each hackathon highlights a different watsonx tech focus like IBM Granite or agentic AI, and a new set of challenges to solve with that tech.
IBM is the founding partner of the Call for Code initiative, now in its 8th year, that calls on problem-solvers around the world to build tech to address social and humanitarian issues. This year Call for Code is integrated into our hackathon series. This means you’ll be able to chose their own adventure within each hackathon; Solve a business use case, or solve a humanitarian issue as part of a Call for Code-sponsored challenge.
Save the date
Registration for our next hackathon, taking place October 23 - November 3, will be opening September 29. Watch for sign-up to go live on the hackathons web site.