ND Labs
ND Labs is a Polish IT company.
How do you ensure an AI agent does not perform unauthorized or crazy transactions?
“Security is our top priority. We implement Programmatic Guardrails and strict risk parameters. You define the agents spending limits, whitelisted smart contracts, and mandatory multi-sig approvals for high-value operations. The agent operates within a sandbox of rules that it cannot bypass, no matter its internal logic.”
在公司資料中查看Can AI agents interact with any blockchain or just specific ones?
“Our agents are chain-agnostic. We build interfaces for EVM-compatible chains (Ethereum, Arbitrum, Base), as well as high-performance networks like Solana and NEAR. Through cross-chain protocols and bridges, we can even develop multi-chain agents that manage liquidity across several ecosystems simultaneously.”
在公司資料中查看How do you prevent LLM "hallucinations" in financial operations?
“We use a Retrieval-Augmented Generation (RAG) architecture combined with "Chain-of-Thought" reasoning. Instead of letting the AI guess, we feed it real-time, verified on-chain data and price oracles. Every decision is cross-referenced with hard data before a transaction is ever proposed or signed.”
在公司資料中查看Do I need to provide my own LLM API keys, or do you host the models?
“We offer flexible deployment. We can integrate with commercial APIs like OpenAI or Anthropic, or deploy open-source models (like Llama 3 or Mistral) on your own private infrastructure or decentralized compute networks like Akash. This ensures total data privacy and eliminates reliance on a single provider.”
在公司資料中查看