📘 Workflow Documentation – Stock Market Daily Digest

👋 Introduction

Wake up to a clean, analyst-style stock digest in your inbox—top gainers/losers, a readable performance table, 3–5 insights, and upcoming events—no spreadsheets, no manual scraping, no copy-paste.
This article explains, step by step, how to build a robust, daily, end-to-end automation that collects market data (Bright Data), waits until scraping is done, aggregates results, asks an AI model (OpenAI) to draft a styled HTML email, logs everything to Airtable, and finally sends the report via Gmail.

You’ll find a friendly but technical tour of every single node, so you can rebuild or adapt the same pipeline with confidence.

---

🎯 Who is this workflow for?

• Investors & traders who want a quick, readable daily summary.
• Finance/Product teams building data-driven alerts/digests.
• Consultants & agencies sending recurring client updates.
• Automation builders prototyping finance ops quickly.

---

🧰 Tools you’ll need

• Bright Data — dataset triggers & snapshots for reliable web data.
• ** OpenAI (GPT)** — to generate a professional HTML digest.
• Airtable — store daily rows for history, filters, dashboards.

Example Airtable Table: Daily Stocks

Ticker	Company	Price	Change %	Sentiment	Date
AAPL	Apple Inc.	225.80	+1.4%	🟢 Positive	2025-09-18 09:00
MSFT	Microsoft Corporation	415.20	-0.7%	🔴 Negative	2025-09-18 09:00
NVDA	NVIDIA Corporation	124.55	+2.1%	🟢 Positive	2025-09-18 09:00
TSLA	Tesla Inc.	260.00	-3.0%	🔴 Negative	2025-09-18 09:00
META	Meta Platforms Inc.	310.45	+0.5%	🟡 Neutral	2025-09-18 09:00

• Gmail — deliver the final HTML email to stakeholders.
• n8n — the automation engine that orchestrates every step.

> Keep API keys in n8n Credentials (never hard-code secrets).

---

🗺️ Architecture at a glance

1. Schedule fires daily
2. Seed list of tickers
3. Split into one item per stock
4. Prepare keyword for scraping
5. Launch Bright Data job
6. Poll progress with a wait-loop
7. Fetch snapshot data
8. Aggregate for the AI
9. Generate HTML summary (GPT)
10. Save rows to Airtable
11. Send email via Gmail

---

⚙️ Step-by-step — Every node explained

⏰ Daily Run Trigger (Schedule Trigger)

Purpose

Start the automation at a precise time each day so nobody needs to push a button.

Parameters (English)

• Trigger Type: Time Interval or Cron
• Every X: 1 Day (or your preferred cadence)
• Timezone: UTC (or your own)
• Start Time: optional (e.g., 09:00)

---

📝 Set Stock List (Set Node – SAMPLE DATA)

Purpose

Define the universe of stocks to monitor. This acts as the seed data for scraping.

Parameters

• Values to Set: Fixed JSON (array of objects)
• Keep Only Set: true
• Fields per item: ticker, name, market_cap (you may add sector, isin, etc.)

---

🔀 Split Stocks (Split Out)

Purpose

Turn the array into individual items so each ticker is processed independently (scraping, polling, results).

Parameters

• Operation: Split Out Items
• Field to Split: the array defined in the previous Set node

---

🏷 Prepare Stock Keyword (Set Node)

Purpose

Create a keyword field (typically equal to ticker) for Bright Data discovery.

Parameters

• Values to Set: Add Field
• Field Name: keyword
• Value: use an expression referencing the current item’s ticker (e.g., {{ $json.ticker }})

---

🕸 Bright Data Scraper (HTTP Request)

Purpose

Trigger the Bright Data dataset to start collecting information for the keyword. Returns a snapshot_id to poll later.

Parameters

• Method: POST
• Endpoint: https://api.brightdata.com/datasets/v1/trigger
• Authentication: Authorization: Bearer <token> (header)
• Body Fields:
  • dataset_id: your Bright Data dataset ID
  • discover_by: usually keyword
  • keyword: the value prepared above

> Add a retry/backoff policy on 429/5xx in node options.

---

🔄 Check Scraper Progress (HTTP Request)

Purpose

Poll Bright Data to see whether the snapshot is running or ready.

Parameters

• Method: GET
• Endpoint: https://api.brightdata.com/datasets/v1/snapshots/{snapshot_id}
• Authentication: Authorization: Bearer <token>
• Expected Output: a status field (running, ready)

---

⏳ Wait for Data (Wait Node)

Purpose

Pause between progress checks to avoid rate limits and give Bright Data time to finish.

Parameters

• Mode: Wait a fixed amount of time
• Time: e.g., 30 seconds (tune to your dataset size)

---

🔀 Scraper Status Switch (Switch Node)

Purpose

Route logic based on the polled status.

Parameters

• Value to Check: status
• Rules:
  • Equals running → go to Wait for Data (then re-check)
  • Equals ready → proceed to Fetch Scraper Results

> Loop pattern: Check → Wait → Check, until ready.

---

📥 Fetch Scraper Results (HTTP Request)

Purpose

Download the completed snapshot data once Bright Data marks it ready.

Parameters

• Method: GET
• Endpoint: https://api.brightdata.com/datasets/v1/snapshots/{snapshot_id}/data
• Authentication: Authorization: Bearer <token>
• Query: format=json
• Output: array of rows per ticker (price, change %, any fields your dataset yields)

> Normalize fields with a Set/Code node if needed.

---

📊 Aggregate Stock Data (Aggregate Node)

Purpose

Combine all individual items into one consolidated object so the AI can analyze the entire market snapshot.

Parameters

• Mode: Aggregate (merge to a single item)
• Fields to Include: ticker, name, price, change, sentiment (plus any extra fields captured)
• Output: one JSON item containing an array/map of the day’s stocks

---

🤖 Generate Daily Summary (AI Node – OpenAI)

Purpose

Ask the model to convert raw data into a styled HTML email: headline, top movers, table, insights, and (optional) upcoming events.

Parameters

• Model: gpt-4.1
• Input: the aggregated JSON from the previous node
• Prompt guidelines:
  • Output HTML only with inline styles (email-safe)
  • Include a table (Ticker, Company, % Change with ↑/↓ & color, Market Cap, Sentiment icon)
  • Highlight top 2 gainers & 2 losers with short reasoning if present
  • Provide 3–5 insights (sector rotation, volatility, outliers)
  • Add upcoming events when available (earnings, launches, macro)
  • Footer: “Generated automatically by your AI-powered stock monitor”
• Output field: confirm the exact property that contains the HTML (e.g., output, message, text)

---

🗂 Save to Airtable (Airtable – Create Record)

Purpose

Log each item (or the roll-up) to Airtable for history, filtering, and dashboards.

Parameters

• Operation: Create Record
• Base ID: from your Airtable URL
• Table: e.g., Daily Stocks
• Field Mapping:
  • Ticker ← {{ $json.ticker }}
  • Company ← {{ $json.name }}
  • Price ← {{ $json.price }}
  • Change % ← {{ $json.change }}
  • Sentiment ← {{ $json.sentiment }}
  • Date ← {{ $now.toISO() }}

> Use a Single-Select for Sentiment (🟢 / 🟡 / 🔴) to build clean Airtable views.

---

📧 Send Report via Gmail (Gmail Node)

Purpose

Deliver the AI-generated HTML digest to your recipients.

Parameters

• Operation: Send Email
• Send To: one or more recipients (e.g., [email protected])
• Subject: Daily Stock Market Digest – {{ $now.format("yyyy-MM-dd") }}
• Message (HTML): reference the AI node’s HTML property (e.g., {{ $('Generate Daily Summary').first().json.output }})
• Options: set Append Attribution to false (keep the email clean)

> Test in Gmail, Outlook, and mobile to validate inline CSS.

---

🧪 Error handling & reliability tips

• Backoff on Bright Data — If scraping many tickers, increase Wait or batch requests.
• Guard against empty results — If a snapshot returns 0 rows, branch to a fallback email (“No data today”).
• AI guardrails — Enforce “HTML-only” and skip missing sections gracefully.
• Airtable normalization — Strip %, cast numbers to float before insert.
• Observability — Add a final Slack/Email On Fail node with run ID and error message.

---

🧩 Customization ideas

• Sector deep-dives: add sector fields and a second AI paragraph on sector rotation.
• CSV attachment: generate & attach a CSV for power users.
• Multiple lists: run parallel branches for Tech, Healthcare, or regions.
• Other asset classes: Crypto, ETFs, Indices, FX.
• Audience targeting: different “To” lists and slightly different prompts per audience.

---

✅ Why this workflow is powerful

• Hands-off — the report simply shows up every day.
• Analyst-grade — clean HTML, top movers, tidy table, actionable insights.
• Auditable — rows archived in Airtable for history and dashboards.
• Composable — swap scrapers, LLMs, storage, or email service.
• Scalable — start with 10 tickers, grow to many lists using the same loop.

For advanced no-code & AI projects, see 0vni – Agence automatisation.