Case studies · Selected engagements
AIKm case studies
Selected AIKm engagements across AI research, retrieval-augmented generation (RAG), chatbots and generative AI. Client names are withheld under NDA; sector and scope are disclosed so readers can judge fit.
Enterprise RAG system for a restaurant and food chain group
Challenge. A multi-brand food and beverage group needed staff and managers to retrieve operating procedures, menu and supplier knowledge on demand.
Our team's approach. We designed and developed a retrieval-augmented generation (RAG) system over the group's own documents, tuned for the group's bilingual operational content.
RAG · Enterprise knowledge · Bilingual
Corporate-wide operations chatbot for an NGO
Challenge. A local NGO needed a single AI assistant that staff across departments could ask about corporate operations, policies and internal procedures.
Our team's approach. We designed and developed a RAG-powered chatbot as an internal operations AI assistant, deployed for corporate-wide use.
RAG · Chatbot · NGO
Construction knowledge retrieval chatbot
Challenge. Practitioners in building and construction needed fast, reliable retrieval across dense technical documentation and standards.
Our team's approach. We designed and developed a chatbot for building and construction knowledge retrieval, indexed over the client's technical knowledge base.
Chatbot · Knowledge retrieval · Construction
AI game role-play system
Challenge. A client needed AI-driven role-play for interactive game characters, with controllable personas and dynamic dialogue.
Our team's approach. We designed and developed an AI game role-play system combining generative dialogue with persona control for in-game characters and scenarios.
Generative AI · Role-play · Games
Deep learning models for 3D applications
Challenge. A local institute needed custom deep learning models to process and interpret 3D data for downstream applications.
Our team's approach. We designed and developed the deep learning architecture, prepared the training pipeline, and delivered a working model integrated into the institute's 3D workflow.
Deep learning · Computer vision · 3D