Hubi, the AI Fair Assistant

aiXbrain creates AI agent Hubi for interactive trade fair guidance

Good partners with commitment and know-how! Thanks to the aiXbrain GmbH team for the great cooperation - for the 2nd time and gladly again in the future

Dr. Alexander Broos

Head of Research and Technology Department Managing Director VDW Research Institute

Scope

The VDW – Association of German Machine Tool Manufacturers is hosting GrindingHub, the leading international trade fair for grinding technology in Stuttgart, featuring around 500 exhibitors from over 30 countries and a multi-day forum program. This is an event that can hardly be fully explored using a traditional catalog and floor plan.

Building on preliminary work for EMO Hannover, the world’s leading trade fair for production technology, the VDW commissioned aiXbrain to develop an AI trade show assistant for grinding technology. AI agent Hubi is designed to guide visitors through the trade show offerings both in preparation and on-site, from searching for exhibitors and personal visit planning to navigation on the day of the event. In doing so, he must rely exclusively on curated, up-to-date trade show data and treat all exhibitors equally. Both are structurally necessary for a trade show organizer. 

Hubi’s tasks are:

  • Research on exhibitors and the forum program
  • Recommendation-based visit planning
  • Orientation and trade show information

Implementation

Hubi is based on an agent-based LLM architecture with specialized skills for research, visit planning, hall navigation, the forum program, and establishing contact. Through dynamic context management, the agent loads only the tools necessary for the current step. This keeps it focused and ensures its responses are precise.

In the background, Hubi combines several AI components: an LLM subagent for translating queries into the trade show context, multiple RAG tools for searching trade show data, algorithmic tools for calculations such as route optimization for booth order, and a restricted web search on the official trade show website. All tools are deliberately limited to curated sources. This design systematically minimizes the risk of hallucinations.

Hubi overcomes the typical readability limitations of traditional chats through structured, interactive interface elements for large result lists. Visitors can use these elements to directly select exhibitors. The selection is fed back to the agent as context and guides the next response.

The system is rounded out by continuous quality control. In-house test suites and real conversations are systematically evaluated to continuously ensure and improve response quality.

Result

Right from the start of the trade show, Hubi was being used extensively by visitors: several hundred conversations with over a thousand inquiries in the first few days of use, spread across all functional areas: exhibitor search, visit planning, navigation, and the forum program.

Exhibitors are found not only by their name, but also by what they offer. Hubi recommends them specifically to visitors who are looking for their topics and solutions with concrete needs. Visitors find not only what they already know, but what suits them. Hubi gives them access to the full scope of GrindingHub in minutes: exhibitors, forum program, and topics. With Hubi, the VDW is leading the way technologically and ensures that exhibitors and visitors at GrindingHub find each other in a targeted manner.

aiXbrain provides Hubi as a managed service: data integration, model operation, frontend, multilingual support, and monitoring all come from a single source.

aiXbrain's blue logo
Subscribe to "aiXbrain Ping" to stay up to date:
Vielen Dank! Ihre Anmeldung wurde erfolgreich übermittelt.
Beim Absenden ist ein Fehler aufgetreten. Bitte versuchen Sie es erneut.
Only meaningful content, always GDPR-compliant. More information under Data Protection.
© 2026 aiXbrain GmbH