Stock watchlist board with one-click deep dive research

I want an app my team opens every morning to actually work our stock watchlist instead of scrolling a spreadsheet. It is a browsing and note taking surface built on our Airtable watchlist table, with live market data from Finnhub layered on top and an on demand research agent sitting behind every ticker.

The main view is a sortable board with one row per ticker. Load the watchlist with the Airtable List Records operation, filtered to rows that are not archived. For each ticker show the live price and day move, both absolute change and percent change, from the Finnhub Quote operation; the company name, industry and logo from Company Profile 2; key valuation metrics from Basic Financials such as market cap, P/E, price to sales, 52 week high and low and beta; and the current analyst split from Recommendation Trends, rendered as a compact strong buy, buy, hold, sell and strong sell bar using the most recent period.

Sort the board by absolute day move descending by default, so the biggest movers in either direction sit on top. Every column header should be clickable to re-sort. Style the day move with semantic up and down colors. Give the last deep dive timestamp its own column so people can see what research is stale, and visually flag any ticker with no deep dive at all or one older than 30 days.

Anyone should be able to add a ticker straight from the board. A small add form takes a symbol, validates it by calling Company Profile 2, and then creates the row with Airtable Create Records, prefilled with the company name and industry that came back. Archiving is a row action that uses Airtable Update Record to set the status field to Archived rather than deleting the row, with a toggle on the board to show archived tickers when someone wants to look back.

Clicking a ticker opens a detail page for that symbol. One side shows recent headlines from Finnhub Company News for roughly the last 14 days, each with its source, date, headline and a link out. The other side is our own running thesis notes. Saving a new note creates a row in a linked notes table with Airtable Create Records, stamped with the author and the time, and also rolls the latest thinking onto the ticker row itself with Airtable Update Record so the notes stay attached to the ticker. Show the note history newest first and do not lose an in progress edit when someone navigates away.

Every ticker also gets a Run deep dive button that kicks off a background agent for that symbol. The agent pulls Company Profile 2, Basic Financials, Company Peers, Company News and Insider Sentiment from Finnhub, then writes a one page research memo covering what the company does, how its valuation compares to the peer group Finnhub returns, what the recent news flow says, and which way insider sentiment has been trending. It files the memo back into that ticker's Airtable record with Update Record along with the run timestamp, so the detail page renders the memo inline underneath the notes and the last deep dive column on the board updates. Show a queued or running state on the button while the agent works, and keep the previous memo visible until the new one lands.

Assume the Airtable watchlist table has fields for the ticker symbol, company name, industry, status of Active or Archived, thesis notes, a long text deep dive memo field rendered as markdown, and a last deep dive timestamp. If one of those fields is missing, tell the user exactly which field to add rather than failing silently.

Bake in the Finnhub free tier limits. Quote, Company Profile 2, Basic Financials, Company News, Recommendation Trends, Company Peers and Insider Sentiment all work on the free plan, but candle and price history endpoints return HTTP 403 without a paid plan, so do not design any price chart that depends on candle data. The free plan is also capped at 60 calls per minute, so fetch quotes for the whole board in a single batched server handler with an explicit refresh control rather than one request per row on every render, and cache the slower moving profile, financials and recommendation data so it is not refetched every time the board loads.