In previous posts, we introduced the PERKS Conversational Chatbot (Solution 6A), its agentic design, its knowledge-graph-grounded retrieval (Graph-RAG), multilingual support, and its extension to spoken interaction through Agentic Voice. Developed by Onlim on its enterprise Conversational AI platform, S6A lets operators, technicians, and engineers consume procedural knowledge through natural-language conversation rather than paging through static manuals.
In this post, we look ahead: how are the capabilities developed in PERKS being carried into real commercial deployments for manufacturing clients?
From Research to Market: Onlim Manufacturing
The capabilities developed through PERKS are not confined to the research lab. Onlim is bringing them to market through Onlim Manufacturing, a dedicated offering for industrial companies and manufacturers. Onlim Manufacturing automates inquiries related to products, service, support, and internal processes, accessible via chat, voice, email, and messenger channels.
The connection to PERKS is direct. The agentic design, knowledge-graph-grounded retrieval, and multichannel delivery developed in the project form the technological backbone of ONLIM Manufacturing. Where PERKS validated these capabilities in research use cases with partners like Beko, Fagor Automation, and Siemens, the commercial offering now extends them to a broader set of manufacturing clients. Two of these clients illustrate the range of applications: Röfix and Romer Labs.
Röfix: Building Materials, Complex Products, High Information Demand
Röfix is an international building materials manufacturer, part of the FIXIT GRUPPE, with roots going back to 1888. The company offers around 200 different building materials, from ready-mixed plasters and mortars to thermal insulation systems, screeds, paints, and coatings, across multiple European markets.
A product portfolio of this breadth generates a high volume of recurring inquiries: Which product is suitable for a given substrate? What are the correct mixing ratios? How should a thermal insulation system be applied step by step? What safety data or technical documentation is available for a specific product?
These are precisely the types of questions that the PERKS Chatbot was designed to handle. By connecting to Röfix’s product knowledge base and technical documentation, the conversational assistant can guide customers, installers, and sales teams through product selection, application procedures, and troubleshooting — in natural language, across multiple channels, and in the languages their European markets require. The agentic design ensures that the conversation can move fluidly between structured procedural guidance (“apply the base coat in this order”) and open-ended question answering (“can I use this product on a damp surface?”).
Romer Labs: Food Safety Diagnostics with Complex Testing Workflows
Romer Labs, a company of dsm-firmenich, is a global leader in diagnostic solutions for food and feed safety. Their product range covers testing for mycotoxins, food allergens, GMOs, microbial contaminants, and more, used by food producers and processors across the entire farm-to-fork supply chain.
For Romer Labs, the challenge is different from Röfix but equally well suited to conversational AI. Laboratory technicians and quality managers need fast, reliable answers to questions about testing protocols: Which test kit is appropriate for a given analyte and sample matrix? What are the correct sample preparation steps? How should results be interpreted against regulatory thresholds? When customers contact Romer Labs’ technical support, they often need procedural guidance that is specific to a product, a testing context, and sometimes a regulatory framework.
The PERKS Chatbot’s strength here lies in its ability to combine structured, step-by-step procedural execution with grounded question answering from product documentation and technical manuals. A user can ask “How do I run an AgraQuant test on corn samples?” and receive a guided walkthrough, while still being able to pause and ask follow-up questions about incubation times or shelf-life conditions. Voice interaction is particularly relevant in laboratory settings where hands may be occupied with sample handling.
A Common Pattern, Different Domains
Röfix and Romer Labs operate in different industries, but the underlying information challenge is remarkably similar: a broad product portfolio, complex procedural knowledge, multilingual and multi-market requirements, and the need to deliver fast, consistent, and reliable answers at scale. Both benefit from a conversational assistant that is grounded in approved knowledge rather than free-floating language model generation, and both need the flexibility to serve users through chat, voice, or messaging depending on the operational context.
This is exactly the scenario that PERKS was designed to address. The research project validated the core capabilities: agentic orchestration, knowledge-graph-grounded retrieval, multichannel delivery, in industrial use cases with Beko, Fagor Automation, and Siemens. Onlim Manufacturing now carries these capabilities forward into commercial deployments, demonstrating that EU-funded research can translate directly into products that solve real operational challenges for manufacturing companies.
Learn More
Read the full description of Solution 6A: “PERKS S6A – Conversational AI for Exploiting Procedural Knowledge”
Explore ONLIM Manufacturing: onlim.com/en/chatbot/solutions/manufacturing/
This work has been supported by the PERKS project (EU Horizon Europe, Grant Agreement No. 101120323).