Use cases/Specialized

Earnings Tracker

Watch earnings calendars + transcripts for stocks you own; get a one-paragraph 'what changed' summary the morning after each report.

Specializedmedium~45m setup
Tools
web_fetchmemorymemory_search
Channels
telegram
Uses
cron

Earnings reports move stocks more than most people's portfolios can handle. Reading every report manually is tedious; reading none means holding through obvious changes. An agent watching only your tickers, summarising only what's material, lands in the right zone.

What it does

  • Tracks tickers you specify (max 10–15 — focus over breadth)
  • Pulls upcoming earnings dates weekly
  • Morning after each report: fetches transcript + press release, summarises what's actually new vs prior guidance
  • Flags surprises (revenue beat/miss > 5%, guide raise/lower, mgmt changes, segment shifts)
  • Daily digest at market close: any tickers down/up >3% with no earnings — pulls news to explain

What you'll need

  • Web fetch for earnings calendars (Yahoo, Seeking Alpha public pages) and transcripts
  • A list of tickers
  • Memory for prior-quarter context
  • Cron

Setup

1. Define your watchlist

Send to Flowly
Remember: Watchlist (with my position size for context, not for advice): - AAPL — large - MSFT — medium - ASML — medium - TSM — small - (add up to 15 total) For each, store: ticker, my-thesis (one paragraph why I own it), last-quarter-key-metrics. Tag: "watchlist".

2. Weekly earnings calendar pull

Send to Flowly
Cron "earnings-calendar" every Sunday at 6 PM: For each ticker in watchlist: 1. Fetch upcoming earnings date (next 30 days) from yahoo finance or similar 2. Store as memory tagged "earnings-upcoming" Send me a Telegram message: 📊 Earnings this week Tue: ASML (premarket) Wed: MSFT (after close) Thu: AAPL (after close)

3. Post-report summary cron

Send to Flowly
Cron "earnings-recap" daily at 7 AM: For each ticker that reported in the last 18 hours: 1. Fetch the earnings press release + first 30 min of conference call 2. Compare against last-quarter-key-metrics in memory: - Revenue, EPS, guidance, segment growth 3. Write a 4-paragraph summary: - Headline: beat/missed/in-line - What's new vs my thesis - Numbers that matter (not all numbers — just material changes) - One-sentence "why this might move the stock" 4. End with: "This is informational, not advice." 5. Send to Telegram, group multiple tickers if multiple reported same day

4. Daily anomaly check

Send to Flowly
Cron "ticker-watch" weekdays at 4:30 PM (after US close): For each ticker, fetch today's % change. For any with |change| > 3% and no earnings in the last 24 hours: 1. Search news for the ticker today 2. Pull top 3 articles, summarise the catalyst in 2 lines 3. Send: "📈 ASML +5.2% today. Catalyst: <summary>"

Tips

  • Limit the watchlist. 15 tickers is the upper bound. Beyond that the morning digest becomes ignorable.
  • Update your thesis quarterly. When earnings shift the story, update your stored thesis. The agent uses it to compare future reports against — stale theses = bad summaries.
  • Don't trade on the digest. Treat it as a "I should now go look at the actual report and form my own view" prompt. The summary is for filtering, not deciding.
  • Watch out for false-precision summaries. AI summaries of earnings calls often invent specific numbers. The agent should QUOTE numbers from the source, not paraphrase them. Bake this into the prompt.
  • This is not financial advice. The agent's outputs are informational. Your broker is for advice.