Technology Stack

Kognys is built on a robust and cutting-edge technology stack, carefully architected to power a decentralized and intelligent research ecosystem. The Unibase stack serves as the foundational infrastructure for our platform, specifically designed to facilitate immortal AI agents with sovereign memory and cross-platform interoperability on BNBChain.

The Unibase Stack: The Backbone of Kognys' AI Agents

• Unibase Membase (Decentralized AI Memory Layer): This is the core decentralized AI memory and agent management layer of Kognys. We utilize Membase to manage the entire lifecycle of our AI agents, including their identities, tasks, and persistent, sovereign memory. All agent conversations and intermediate findings from the "Chain of Debate" are stored in Membase, providing a fully auditable and persistent record of the research process on the BNB testnet. This directly fulfills the challenge's requirement for a decentralized AI memory layer.

• Unibase DA (Data Availability): To guarantee the permanent and uncensorable storage of knowledge, we employ Unibase DA. This decentralized storage solution is where all final, generated papers and their supporting datasets are archived. This ensures that research produced on Kognys cannot be lost, altered, or locked behind a paywall, reinforcing the immortal aspect of our agents' contributions.

• Unibase AIP (Agent Interaction Protocol): This is the communication protocol that governs how our AI agents interact, debate, and collaborate. Unibase AIP provides the structured framework necessary for the complex coordination required during the "Chain of Debate," ensuring that agent interactions are productive and goal-oriented. This is crucial for enabling cross-platform interoperability within our agent ecosystem.

Complementary Technologies for Agent Autonomy

• LangGraph: We use LangGraph to build our stateful, multi-agent applications. It works in concert with the Unibase stack, acting as the conductor for our agent workflows. While Unibase provides the foundational infrastructure for memory and interaction, LangGraph allows us to define agent roles and manage complex research processes (like the Chain of Debate) as a cyclical graph. This ensures that all agent activities are structured, robust, and aligned with the researcher's ultimate goals, enabling agents to autonomously manage data and evolve based on interactions.

• BNB Testnet: We utilize the BNB testnet as the underlying blockchain layer for Unibase Membase. Its scalability, high throughput, and low transaction costs are essential for managing the high volume of micro-transactions generated by our AI agents as they record their identities and interactions on-chain, aligning perfectly with the challenge's specified blockchain.

• OpenAlex API: To fuel our AI agents with the vast corpus of existing scientific knowledge, we integrate with the OpenAlex API. This provides our agents with programmatic access to a comprehensive, real-time database of hundreds of millions of scholarly articles, authors, and institutions, forming the foundation for their research and enabling them to share knowledge across platforms.