Dnotitia

What Dnotitia does

Dnotitia is a pioneering AI company founded in 2023, specializing in integrating AI and data to drive innovative value. They develop Seahorse, a high-performance vector database powered by proprietary Vector Data Processing Units (VDPU), and Mnemos, a cutting-edge service solution for Large Language Models (LLMs) and an advanced edge AI device. Dnotitia also leads the development of Small Language Models (SLMs) and Retrieval-Augmented Generation (RAG) systems, positioning itself at the forefront of GenAI innovation. Their offerings include a cloud-native vector database platform, portable LLM devices, and a bilingual foundation model optimized for Korean and English.

Products

• SeahorseA high-performance vector database powered by proprietary Vector Data Processing Units (VDPU), designed to process large-scale and complex data more effectively. It offers multimodal semantic search, exceptional speed, cost efficiency, high accuracy, and advanced filtering. Seahorse Cloud is a comprehensive cloud-native RAG solution platform with AI model hosting, no-code pipeline design tools, and automated optimization capabilities.
• MnemosA cutting-edge service solution for Large Language Models (LLMs) and an advanced edge AI device for personal LLM use. It enables running high-performance LLMs anywhere without data centers or cloud infrastructure, leveraging advanced optimization and LLM technologies for scalable, cost-efficient, real-time, on-premise AI deployments. It is built on hardware optimized for LLMs and powered by advanced SLM technology.
• DNA: Dnotitia AI Foundation ModelA state-of-the-art bilingual AI model built on the Llama architecture with 8 billion parameters, optimized for understanding and generating content in both Korean and English with unmatched precision. It is available on Hugging Face as an open-source model.
• AI ConsoleA secure AI system ensuring data privacy.

Key capabilities

• Proprietary Vector Data Processing Units (VDPU)
• High-performance vector database
• Multimodal semantic search
• LLM optimization and compression
• Compatible SLM foundation model
• Low-power, high-performance edge AI
• Automatic tuning and deployment
• Cloud-native vector database platform
• End-to-End RAG pipeline solution
• No-code pipeline design tools
• Intelligent optimization (AutoRAG, batch processing)
• Diverse storage integration
• Microservices architecture
• Containerized microservices structure
• Kubernetes orchestration
• Multi-cloud support (AWS, Azure, GCP)
• Enterprise security (data encryption, granular access controls, SSO, audit logging)
• Auto-scaling infrastructure
• Bilingual AI model (Korean and English)
• Llama architecture based (8 billion parameters)
• Knowledge distillation
• Continual pre-training with high-quality Korean dataset
• Supervised fine-tuning (SFT)
• Direct preference optimization (DPO)

Use cases

• Transforming multimodal data into precise search results
• Running high-performance LLMs on edge devices
• Enhancing LLM quality through Retrieval Augmented Generation (RAG)
• Advanced semantic search for unstructured data
• Building high-performance LLM applications
• Real-time, on-premise AI deployments
• Understanding and generating content in Korean and English
• Medical AI for improved diagnostic and treatment accuracy
• Insurance sales market transformation (AI Planner solution)
• LLMOps platform enhancement

AI approach

Dnotitia develops proprietary AI solutions including Seahorse, a high-performance vector database powered by its proprietary Vector Data Processing Units (VDPU), and Mnemos, an edge AI device for personal LLMs. They also develop Small Language Models (SLMs) and Retrieval-Augmented Generation (RAG) systems. Their DNA: Dnotitia AI foundation model is a bilingual model (Korean and English) built on the Llama architecture, optimized for understanding and generating content with high precision, and is available on Hugging Face as open-source. They also have an advanced Vision Language Model technology.

Tech named: Vector Database, LLM, SLM, Retrieval-Augmented Generation (RAG), Vector Data Processing Units (VDPU), LLM Optimization & Compression, AI embeddings, AutoRAG, Llama architecture, spherical linear interpolation (SLERP), knowledge distillation (KD), continual pre-training (CPT), supervised fine-tuning (SFT), direct preference optimization (DPO), Vision Language Model (VLM)

Industries served

• IT System Design Services
• Insurance
• Healthcare
• Medical Imaging
• Smart Factory
• Government
• Defense

What it says sets it apart

• Pioneering AI company integrating AI and data
• Proprietary Vector Data Processing Units (VDPU)
• First integrated hardware and software LLM solution for maximum efficiency
• World's first hardware-accelerated vector database
• Advanced Vision Language Model Technology
• Affordable LLM inferencing system at scale
• Unmatched precision in understanding Korean culture and context for their DNA model
• Exceptional benchmark results in KMMLU for Korean AI
• Open-source availability of DNA model on Hugging Face

Funding rounds we track

From the AI funding tracker — rounds as reported by the linked publications.