Lumen is a concept for an AI research assistant built for investors and traders. You ask a plain-language question about a company, and Lumen reads its filings, earnings calls and news, then answers with every claim linked to its source. I designed the marketing site, app dashboard, pricing page, onboarding, mobile answer and alert screens, and a set of launch visuals.

Investors work across filings, transcripts and news that live in different places. Finding out why a stock moved means opening many tabs and cross-checking by hand, usually under time pressure.

AI can compress that work, but finance is unforgiving. A confident answer that can’t be checked is worse than no answer, because a wrong claim can cost real money. The challenge was to design an AI tool that is fast enough to be useful and transparent enough to be trusted.

Make every claim verifiable. Each statement should be one tap from its source.

Reduce research time. Get from a question to a usable answer on one screen.

Signal technical credibility. The product should look and feel like serious infrastructure, not a chatbot wrapper.

I started by reviewing existing research and trading tools and reading about how investors verify information. Three themes shaped the design. These are working hypotheses to validate with interviews.

Trust depends on traceability. Investors are skeptical by nature. An answer is only useful if they can check where it came from in seconds.

Information is fragmented. The value is in reading filings, calls and news together and connecting the dots.

Technical buyers want to see how it works. For SaaS and AI products, showing the process builds more confidence than polished marketing copy.

Citations as a core element. Every answer includes numbered source chips for filings, calls and news. A user can see at a glance how well-supported a claim is. It is the single most important trust signal in the product.

A terminal-style answer console. The hero shows a real query and answer in a window styled like a developer tool. It communicates the product’s purpose in seconds and signals a technical audience.

A visible pipeline. The landing page explains the process in four steps: Ingest, Index, Reason and Cite. Showing how the answer is built makes the product feel auditable instead of magical.

An API section. A code block shows a request and response, with the sources included in the response. It tells technical buyers that the product can plug into their own stack.

Restrained use of color. Because green and red only ever mean “up” and “down,” traders can read the screen without translating it.

A clear disclaimer. A short “research tool, not financial advice” line sits on screens that show market data. It’s a small detail, but it shows an understanding of the domain.

Lumen shows how an AI product in a high-stakes field can earn trust through design. Citations sit at the center of every screen, the pipeline and API make the product feel inspectable, and a restrained palette keeps the interface calm when the data is not. The result is a consistent system across desktop, mobile and launch visuals.