Fina: a voice financial research assistant, powered by Vocal Bridge

A hands-free market research agent (its assistant is named Finn) that takes layered, multi-part questions out loud and answers with a synthesized verbal briefing, built on Vocal Bridge's Voice for App surface.

The problem: market research is too many steps for the question in your head

Financial research is a juggling act. To answer one real question (how exposed are these three names to the current macro environment), you pull price history from one place, volatility and beta from another, news from a third, then read it all and form a view. Every follow-up means re-filtering and re-querying. The question in an investor's head is conversational and layered. The tooling is a stack of dashboards and search boxes that each expect a narrow, structured input.

That gap is worst exactly when someone wants to think out loud. A dashboard makes you decide the timeframe, the tickers, and the metric before you can ask anything. A conversation lets you start broad and narrow as you go.

What Fina built

Fina turns market research into a spoken conversation. You ask a broad, multi-part question and get a synthesized briefing back, then keep going with follow-ups that build on what was already said, with no restating of context. The assistant gathers the data, reads back a structured summary, and stays in the thread as you pivot from one company or topic to the next.

The point is not any single number. It is that a person can hold a real, compound research conversation hands-free and the agent keeps the thread. Voice fits because layered market questions are easier to ask out loud than to type, click, and re-filter one at a time.

How Fina uses Vocal Bridge

Fina is built on Vocal Bridge's Voice for App surface, where the voice agent and a browser UI share the same channel. The page shows a "Start Voice Chat" control, a "Connected, ready to chat" status, animated volume bars, and an "Agent is speaking" cue while a request is processing. The voice agent drives that state, and the UI reflects what the agent is doing.

The architectural split is clean. Fina owns the financial side, market data retrieval, the volatility and beta calculations, news aggregation, and the analysis that turns raw figures into a verbal briefing. Vocal Bridge owns the real-time voice channel, the streaming text-to-speech, barge-in so the user can interrupt, transcript streaming, and the conversational filler phrases that hold a session together while a long-running query completes.

import { useAgentActions } from "@vocalbridgeai/react";

const { onAction, sendAction } = useAgentActions();

// agent -> app: reflect the current research step in the UI
onAction("research_status", ({ stage }) => {
  setStatus(stage); // e.g. "analyzing", "ready"
});

// app -> agent: user picked a ticker from the sidebar, tell the agent
const handleSelectTicker = (symbol) => sendAction("focus_ticker", { symbol });

Illustrative. The actual Fina implementation carries richer financial entities and tool calls.

What's hard about voice here

Real financial analysis is not instant. A compound question can need several data pulls before there is anything to say, and that gap cannot read as broken. Vocal Bridge's streaming text-to-speech, barge-in, and conversational fillers keep the session feeling alive while a slow lookup runs, so the user stays in the conversation instead of staring at silence.

Making the analysis fast enough that the filler is rarely needed is Fina's own work. Keeping the conversation graceful while it runs is the layer Vocal Bridge handles.