Tokenic&Fin

Tokenic&Fin is a next-generation banking ecosystem designed around two core pillars: a user-support chatbot and asset tokenization. At its heart lies a RAG (Retrieval-Augmented Generation) model: internal banking documents are embedded into a vector store, retrieved at query time to enrich responses, and combined with large-language model generation to ensure both accuracy and domain-specific grounding. This reduces hallucinations and ensures answers are up-to-date with bank-specific policy, asset data, and regulatory constraints.

To build and scale this, the system uses Strands Agents SDK together with Amazon Bedrock AgentCore. Strands provides the open-source framework for defining modular agents, tools, memory, and hierarchical agent structures (supervisor + specialist child agents), while Bedrock AgentCore offers secure, serverless runtime with session isolation, identity integration (e.g. Cognito), observability, and scalability.

A particularly novel component is the Smart Contract Assistant and reputation verification layer. When users create encrypted data transfer contracts or interact with tokenization features, this assistant helps with form filling, checks reputation and data correctness, and ensures that smart contracts are finally reviewed by managers before going live.