PROJECT 03 / AI agents · Financial research
Te Vārta.
Designing an investment research assistant that remembers the context behind the question.
Independent product development
Builder
In development
The problem
Fundamental research is an ongoing process: interests, prior questions, and relevant sources evolve over time. Te Vārta is being designed to preserve that context rather than treating every question as a fresh start.
Design direction
Exploring user-specific knowledge graphs, memory, financial information retrieval, and source provenance. The aim is to make research more personalized and contextual over time.
Technical exploration
Retrieval-augmented generation, information retrieval, and agent-driven workflows are areas under exploration. Personalized research briefs and contextual summaries are proposed product directions, not claims of live functionality.
Current progress
The project is in development. Work is focused on designing the research assistant and investigating how context, memory, and attributable information could fit together.
Limitations & next questions
Reliable financial retrieval, source freshness, conflicting evidence, and control over personal research memory remain design questions. The assistant is a research concept and does not establish investment performance.