Dnotitia Open-Sources AKB to Give Enterprise AI a Long-Term Memory

SEOUL, South Korea – May 20, 2026 – Dnotitia Inc., a rapidly growing AI infrastructure company, today announced the open-sourcing of its Agent Knowledge Base (AKB) on GitHub. The platform is designed to provide AI agents with a form of 'long-term memory,' moving beyond current technologies to create a unified, dynamic, and secure knowledge infrastructure for enterprises.

As organizations increasingly integrate generative AI into core workflows, they face a significant hurdle: critical business knowledge is often fragmented across countless documents, databases, chat logs, and applications. Dnotitia's AKB aims to solve this by creating an AI-ready knowledge layer that not only consolidates this scattered information but also continuously learns from the work AI agents perform.

Beyond Retrieval: A New Kind of AI Memory

For the past few years, Retrieval-Augmented Generation (RAG) has been the standard for giving large language models access to external information. However, traditional RAG systems primarily function as sophisticated search engines for static documents. Dnotitia's AKB is engineered to represent a significant leap forward.

Its key differentiator is the ability to continuously accumulate and structure dynamic context. This includes the conversations, work records, decision-making rationales, and outputs generated by AI agents as they execute tasks. Instead of just retrieving a document, an AI agent using AKB can understand the history of a project, the context of a decision, and the relationships between different pieces of information.

At its core, AKB employs an ontology-based structure that defines semantic relationships between data. This allows an AI to grasp not just individual facts but the web of connections that constitute true organizational knowledge. This sophisticated framework integrates various content types, from standard Markdown documents and SQL databases to object storage, creating a single source of truth that both humans and AI agents can leverage.

This system is powered by Dnotitia's proprietary Seahorse vector database, which facilitates advanced semantic search capabilities. The combination enables users and agents to explore business context through graph-based relationships, uncovering insights that would be nearly impossible to find with simple keyword searches.

Dnotitia's Strategic Play in the AI Infrastructure Race

The decision to open-source AKB is not merely a technical release but a calculated strategic move by a company with significant momentum. Founded in 2023, Dnotitia has quickly amassed $76.3 million in funding, including a recent $61.2 million Series A round led by Elohim Partners, signaling strong investor confidence in its vision.

That vision extends beyond a single software product. Dnotitia is building a comprehensive 'AI Storage' stack designed to address data and memory bottlenecks from the ground up. This portfolio includes the Seahorse vector database and, most ambitiously, a custom semiconductor chip called the Vector Data Processing Unit (VDPU). The VDPU, expected to be commercialized in 2027, is designed to dramatically accelerate vector database operations, promising up to 10x faster search speeds and an 80% reduction in total cost of ownership compared to CPU-based systems.

By open-sourcing AKB, Dnotitia is employing a classic ecosystem-building strategy. It aims to establish AKB as a foundational layer for enterprise AI, encouraging widespread adoption and community-driven improvement. This positions the company's commercial offerings, like the high-performance Seahorse database and the VDPU hardware, as natural, powerful upgrades for organizations that build their AI workflows on the open-source AKB framework.

An Open-Source Gambit in a Crowded Field

Dnotitia's move comes amid an 'open-source AI renaissance.' A 2025 survey by McKinsey and the Mozilla Foundation found that 76% of technology leaders plan to increase their use of open-source AI, seeking flexibility, transparency, and freedom from vendor lock-in. The market for AI knowledge management is already bustling with tools like Atlassian Confluence, Guru, and Glean, while specialized agent memory systems like GBrain and Mem0 are also gaining traction.

However, AKB is tailored to carve out a specific, high-value niche. While many open-source agent memory projects are designed for single operators or developers, AKB is explicitly built for complex, multi-tenant enterprise environments. Its architecture anticipates the challenges of deploying AI agents at scale across an entire organization, a key differentiator from many existing solutions.

Balancing Open Access with Ironclad Security

Perhaps the most critical feature for its target audience is AKB's emphasis on security and governance. For many Chief Information Security Officers (CISOs), the idea of autonomous AI agents accessing a wide range of company data is a significant concern. AKB addresses this head-on with a robust, built-in access control system.

The platform incorporates granular permissions at the organization, team, role, and even individual user level. This ensures that AI agents operate within strictly defined boundaries, accessing only the business context they are permitted to see. This ability to balance powerful knowledge sharing with stringent enterprise security requirements is crucial for adoption in highly regulated industries such as finance, healthcare, and legal services. By designing for security from the core, Dnotitia is removing a major barrier that has slowed the deployment of collaborative AI in sensitive corporate settings.

"Enterprise AI competitiveness is increasingly shifting from which models a company adopts to how effectively AI can use the data and knowledge the organization already has," said MK Chung, CEO of Dnotitia, in the company's press release. "By open-sourcing AKB, Dnotitia aims to help more organizations turn their internal knowledge into AI-ready assets and grow alongside with AI agents."

With its non-commercial license available for free on GitHub, Dnotitia is inviting developers and businesses to build upon its platform. The company's strategy suggests a future where the true power of enterprise AI lies not in the model itself, but in the intelligence of the memory it can access. By providing the tools to build that memory, Dnotitia is making a bold bid to become an indispensable part of the next generation of artificial intelligence.