LangCloud Lab

LangCloud Lab is where we share our research and development and open-source activities. We pursue foundational research aimed at real-world AI applications alongside open-source software that anyone can use.

Focus Areas in R&D

Focusing on the reliable application of advanced AI in real-world scenarios, we conduct research in the following areas.

Real-World AI Agent Development

We are developing "Gemba AI" (codename), which provides a "place for human-AI collaboration" across diverse on-site settings to drive AX/DX transformation in the field.

・Gemba AI Development Project

Exploring Higher-Order Circuits in Large Language Models

As part of our mechanistic interpretability research, we are working to elucidate how higher-order circuits are acquired during model training and how they emerge during inference. (Research areas: AI safety, LLM foundational theory, computational study of cognitive function)

・Toward White-Boxing the Internal Circuits of LLMs ・Searching for Metaphor Circuits within LLMs ・BabyLM Research

Multimodal AI Applications for Workspaces and Living Spaces

We research and develop technology that captures real-world workspaces and living spaces to achieve safety, automation, and optimization.

・Easy Visual Data Organization & Utilization App Development

Digital Twins of Society and Economy

We research and develop AX technology that models society and economic activity as a whole to realize safe, vibrant local communities.

Open Source Projects

LLM Lab for Smart Factory

AI Colab Worker® Series

Open-source software for building generative AI applications for manufacturing. Combine RAG, multimodal LLMs, and OCR to build GenAI apps tailored to on-site operations.

View on GitHub