mirror of
https://github.com/khoaliber/n8nworkflows.xyz.git
synced 2026-04-19 17:14:37 +00:00
creation
This commit is contained in:
nusquama
@@ -0,0 +1,290 @@
|
||||
Send weekly AI summaries of Google Sheets data via Gmail using OpenAI
|
||||
|
||||
https://n8nworkflows.xyz/workflows/send-weekly-ai-summaries-of-google-sheets-data-via-gmail-using-openai-13404
|
||||
|
||||
|
||||
# Send weekly AI summaries of Google Sheets data via Gmail using OpenAI
|
||||
|
||||
## 1. Workflow Overview
|
||||
|
||||
**Purpose:**
|
||||
This workflow automatically generates a **weekly business-style summary** from a Google Sheets dataset using **OpenAI (Chat model)**, sends the summary by **Gmail**, and then **logs** the send event into a separate Google Sheet (“Report Log”).
|
||||
|
||||
**Primary use cases:**
|
||||
- Weekly KPI/status email from a spreadsheet (sales, support tickets, marketing metrics, ops metrics)
|
||||
- Automated stakeholder reporting without manual copy/paste
|
||||
- Maintaining an audit trail of sent reports
|
||||
|
||||
### 1.1 Schedule & Data Extraction
|
||||
Runs on a weekly schedule (Monday 9:00), reads all rows from a specified Google Sheet, and aggregates them into one payload.
|
||||
|
||||
### 1.2 AI Summarization
|
||||
Sends the aggregated rows to an AI Agent powered by an OpenAI chat model, with instructions to produce a concise, bullet-point weekly report.
|
||||
|
||||
### 1.3 Formatting, Email Delivery & Logging
|
||||
Formats the AI output into email-ready fields (subject/date/body), sends the email via Gmail, then appends a log entry to a “Report Log” sheet.
|
||||
|
||||
---
|
||||
|
||||
## 2. Block-by-Block Analysis
|
||||
|
||||
### Block 1 — Schedule & Read Data
|
||||
**Overview:**
|
||||
Triggers weekly, pulls spreadsheet rows from Google Sheets, then consolidates all row items into a single JSON field for downstream AI processing.
|
||||
|
||||
**Nodes involved:**
|
||||
- Schedule Trigger
|
||||
- Read Google Sheets Data
|
||||
- Aggregate Rows
|
||||
|
||||
#### Node: Schedule Trigger
|
||||
- **Type / role:** `n8n-nodes-base.scheduleTrigger` — time-based entry point.
|
||||
- **Configuration (interpreted):**
|
||||
- Runs **every week** on **Monday** at **09:00** (server/workflow timezone applies).
|
||||
- **Inputs/outputs:**
|
||||
- **Output →** Read Google Sheets Data (main)
|
||||
- **Version notes:** typeVersion **1.2**
|
||||
- **Edge cases / failures:**
|
||||
- Timezone mismatch (n8n instance timezone vs. expected business timezone).
|
||||
- If workflow is inactive (`"active": false`), it will never run.
|
||||
|
||||
#### Node: Read Google Sheets Data
|
||||
- **Type / role:** `n8n-nodes-base.googleSheets` — reads rows from a sheet.
|
||||
- **Configuration (interpreted):**
|
||||
- Uses a **Document ID** placeholder: `<Google Sheet ID with your data>`
|
||||
- Uses a **Sheet name** placeholder: `<Sheet name e.g. Sales Data>`
|
||||
- Operation is implied as “read/get rows” (default for this node when not explicitly set in parameters).
|
||||
- **Key variables/expressions:** none (placeholders must be replaced).
|
||||
- **Inputs/outputs:**
|
||||
- **Input ←** Schedule Trigger
|
||||
- **Output →** Aggregate Rows
|
||||
- **Version notes:** typeVersion **4.7**
|
||||
- **Edge cases / failures:**
|
||||
- Google auth/permissions (403) if the connected account lacks access.
|
||||
- Wrong sheet name or document ID → “not found”.
|
||||
- Large sheets may cause performance issues or pagination/limits depending on node settings.
|
||||
|
||||
#### Node: Aggregate Rows
|
||||
- **Type / role:** `n8n-nodes-base.aggregate` — combines multiple incoming items into one item.
|
||||
- **Configuration (interpreted):**
|
||||
- Mode: **Aggregate All Item Data**
|
||||
- Stores the combined dataset in field: **`all_rows`**
|
||||
- **Inputs/outputs:**
|
||||
- **Input ←** Read Google Sheets Data
|
||||
- **Output →** Generate Summary with AI
|
||||
- **Version notes:** typeVersion **1**
|
||||
- **Edge cases / failures:**
|
||||
- Empty sheet (0 rows) → downstream AI gets `[]` which may produce an unhelpful summary unless handled.
|
||||
- Very large combined payload can exceed AI model input limits.
|
||||
|
||||
---
|
||||
|
||||
### Block 2 — AI Summarization (OpenAI via LangChain Agent)
|
||||
**Overview:**
|
||||
Takes the aggregated rows and prompts an AI agent to produce a short, professional weekly report (bullet points, <300 words, no greeting/sign-off).
|
||||
|
||||
**Nodes involved:**
|
||||
- Generate Summary with AI
|
||||
- OpenAI Chat Model
|
||||
|
||||
#### Node: OpenAI Chat Model
|
||||
- **Type / role:** `@n8n/n8n-nodes-langchain.lmChatOpenAi` — provides the language model to the agent.
|
||||
- **Configuration (interpreted):**
|
||||
- Model: **`gpt-4.1-mini`**
|
||||
- No special options or built-in tools configured.
|
||||
- **Inputs/outputs:**
|
||||
- Connected via **AI language model port** to Generate Summary with AI.
|
||||
- **Version notes:** typeVersion **1.3**
|
||||
- **Edge cases / failures:**
|
||||
- Missing/invalid OpenAI credentials.
|
||||
- Model unavailable in the account/region.
|
||||
- Rate limits or transient 5xx errors.
|
||||
- Token/context overflow if `all_rows` is too large.
|
||||
|
||||
#### Node: Generate Summary with AI
|
||||
- **Type / role:** `@n8n/n8n-nodes-langchain.agent` — agent that generates the summary text.
|
||||
- **Configuration (interpreted):**
|
||||
- Prompt type: **Define** (custom prompt)
|
||||
- **User prompt** (key expression):
|
||||
- Embeds aggregated data: `{{ JSON.stringify($json.all_rows) }}`
|
||||
- Asks for “brief weekly summary report highlighting key trends, totals, and notable changes… formatted as clean email body with bullet points.”
|
||||
- **System message**:
|
||||
- “Business analyst assistant”, concise/professional, bullet points, **under 300 words**, **no greetings/sign-offs**.
|
||||
- **Inputs/outputs:**
|
||||
- **Input ←** Aggregate Rows
|
||||
- **AI Model input ←** OpenAI Chat Model (ai_languageModel connection)
|
||||
- **Output →** Format Report
|
||||
- **Version notes:** typeVersion **3**
|
||||
- **Edge cases / failures:**
|
||||
- If `all_rows` contains complex/large objects, `JSON.stringify` can bloat the prompt.
|
||||
- If sheet columns are inconsistent, the model may misinterpret data types.
|
||||
- Output format variability: the next node expects `{{$json.output}}` to exist.
|
||||
|
||||
---
|
||||
|
||||
### Block 3 — Format, Email, Log
|
||||
**Overview:**
|
||||
Normalizes the AI output into predictable fields, sends the email via Gmail, then appends a row to a logging spreadsheet.
|
||||
|
||||
**Nodes involved:**
|
||||
- Format Report
|
||||
- Send Report Email
|
||||
- Log Report Sent
|
||||
|
||||
#### Node: Format Report
|
||||
- **Type / role:** `n8n-nodes-base.set` — creates/overwrites fields for downstream usage.
|
||||
- **Configuration (interpreted):**
|
||||
- Creates fields:
|
||||
- `report_body` = `{{ $json.output }}`
|
||||
- `report_date` = `{{ $now.format('yyyy-MM-dd') }}`
|
||||
- `subject_line` = `Weekly Report - {{ $now.format('yyyy-MM-dd') }}`
|
||||
- **Inputs/outputs:**
|
||||
- **Input ←** Generate Summary with AI
|
||||
- **Output →** Send Report Email
|
||||
- **Version notes:** typeVersion **3.4**
|
||||
- **Edge cases / failures:**
|
||||
- If the Agent node output key is not `output` (varies by configuration/version), `report_body` becomes empty/undefined.
|
||||
- Date formatting uses `$now`; timezone can affect the date stamp near midnight boundaries.
|
||||
|
||||
#### Node: Send Report Email
|
||||
- **Type / role:** `n8n-nodes-base.gmail` — sends an email through Gmail OAuth2.
|
||||
- **Configuration (interpreted):**
|
||||
- To: placeholder `<Recipient email address>`
|
||||
- Subject: `{{ $json.subject_line }}`
|
||||
- Message body: `{{ $json.report_body }}`
|
||||
- Credentials: **Gmail OAuth2** (named “Gmail account” in the workflow)
|
||||
- **Inputs/outputs:**
|
||||
- **Input ←** Format Report
|
||||
- **Output →** Log Report Sent
|
||||
- **Version notes:** typeVersion **2.1**
|
||||
- **Edge cases / failures:**
|
||||
- OAuth token expired/revoked → authentication failure.
|
||||
- Gmail sending limits/quota issues.
|
||||
- If `report_body` is empty, email will be blank unless validated.
|
||||
|
||||
#### Node: Log Report Sent
|
||||
- **Type / role:** `n8n-nodes-base.googleSheets` — appends a row to a log sheet.
|
||||
- **Configuration (interpreted):**
|
||||
- Operation: **Append**
|
||||
- Document ID placeholder: `<Google Sheet ID for report log>`
|
||||
- Sheet name: **Report Log**
|
||||
- Columns mapped (define below):
|
||||
- `date` = `{{ $json.report_date }}`
|
||||
- `sent_at` = `{{ $now.toISO() }}`
|
||||
- `subject` = `{{ $json.subject_line }}`
|
||||
- **Inputs/outputs:**
|
||||
- **Input ←** Send Report Email
|
||||
- No downstream nodes.
|
||||
- **Version notes:** typeVersion **4.7**
|
||||
- **Edge cases / failures:**
|
||||
- “Report Log” sheet missing or columns don’t match expected headers.
|
||||
- Permission/auth issues on the logging spreadsheet.
|
||||
- Partial failure risk: email could send successfully but logging fails (no compensating handling in this workflow).
|
||||
|
||||
---
|
||||
|
||||
## 3. Summary Table
|
||||
|
||||
| Node Name | Node Type | Functional Role | Input Node(s) | Output Node(s) | Sticky Note |
|
||||
|---|---|---|---|---|---|
|
||||
| Schedule Trigger | n8n-nodes-base.scheduleTrigger | Weekly trigger (Mon 09:00) | — | Read Google Sheets Data | ## Schedule & Read<br>Runs every Monday at 9am, reads all rows from the data spreadsheet, and aggregates them. |
|
||||
| Read Google Sheets Data | n8n-nodes-base.googleSheets | Read source spreadsheet rows | Schedule Trigger | Aggregate Rows | ## Schedule & Read<br>Runs every Monday at 9am, reads all rows from the data spreadsheet, and aggregates them. |
|
||||
| Aggregate Rows | n8n-nodes-base.aggregate | Combine all rows into one payload (`all_rows`) | Read Google Sheets Data | Generate Summary with AI | ## Schedule & Read<br>Runs every Monday at 9am, reads all rows from the data spreadsheet, and aggregates them. |
|
||||
| OpenAI Chat Model | @n8n/n8n-nodes-langchain.lmChatOpenAi | Provides OpenAI chat model to agent | — (AI port connection only) | Generate Summary with AI (ai_languageModel) | ## Summarize, Send & Log<br>AI generates a summary report, emails it via Gmail, and logs the send to a separate spreadsheet. |
|
||||
| Generate Summary with AI | @n8n/n8n-nodes-langchain.agent | Generate weekly summary from aggregated rows | Aggregate Rows + OpenAI Chat Model (AI port) | Format Report | ## Summarize, Send & Log<br>AI generates a summary report, emails it via Gmail, and logs the send to a separate spreadsheet. |
|
||||
| Format Report | n8n-nodes-base.set | Build report fields (body/date/subject) | Generate Summary with AI | Send Report Email | ## Summarize, Send & Log<br>AI generates a summary report, emails it via Gmail, and logs the send to a separate spreadsheet. |
|
||||
| Send Report Email | n8n-nodes-base.gmail | Send report via Gmail | Format Report | Log Report Sent | ## Summarize, Send & Log<br>AI generates a summary report, emails it via Gmail, and logs the send to a separate spreadsheet. |
|
||||
| Log Report Sent | n8n-nodes-base.googleSheets | Append log row (date/sent_at/subject) | Send Report Email | — | ## Summarize, Send & Log<br>AI generates a summary report, emails it via Gmail, and logs the send to a separate spreadsheet. |
|
||||
| Sticky Note | n8n-nodes-base.stickyNote | Documentation note (not executed) | — | — | ## How it works<br>This workflow runs automatically every Monday at 9am. It reads data from a Google Sheets spreadsheet, aggregates all the rows into a single payload, and sends it to OpenAI to generate a concise weekly summary report with key trends and highlights. The AI-generated report is formatted with a date-stamped subject line, emailed to your chosen recipient via Gmail, and logged to a separate spreadsheet so you have a record of every report sent.<br><br>## Setup steps<br>1. Adjust the Schedule Trigger if you want a different day or time<br>2. Connect your Google Sheets account and set the spreadsheet ID and sheet name containing your data<br>3. Connect your OpenAI credentials to the OpenAI Chat Model node<br>4. Set the recipient email address in the Send Report Email node<br>5. Connect your Gmail account for sending the report<br>6. Create a "Report Log" sheet with columns: date, subject, sent_at<br>7. Set the Report Log spreadsheet ID in the Log Report Sent node<br>8. Activate the workflow |
|
||||
| Sticky Note1 | n8n-nodes-base.stickyNote | Documentation note (not executed) | — | — | |
|
||||
| Sticky Note2 | n8n-nodes-base.stickyNote | Documentation note (not executed) | — | — | |
|
||||
|
||||
> Note: Sticky notes are non-executing annotation nodes; they appear as rows above for completeness because they are nodes in the workflow JSON.
|
||||
|
||||
---
|
||||
|
||||
## 4. Reproducing the Workflow from Scratch
|
||||
|
||||
1) **Create a new workflow** in n8n
|
||||
- Name it similar to: *Send weekly AI-summarized Google Sheets reports via Gmail using OpenAI*
|
||||
|
||||
2) **Add “Schedule Trigger” (Schedule Trigger node)**
|
||||
- Set interval to **Weeks**
|
||||
- Choose **Monday**
|
||||
- Set time to **09:00**
|
||||
- (Optional) Confirm n8n instance timezone aligns with your reporting timezone
|
||||
|
||||
3) **Add “Read Google Sheets Data” (Google Sheets node)**
|
||||
- Connect **Schedule Trigger → Read Google Sheets Data**
|
||||
- Authenticate Google Sheets (Google OAuth2 in n8n)
|
||||
- Set:
|
||||
- **Document**: your source Google Sheet ID
|
||||
- **Sheet name**: the tab containing your data (e.g., “Sales Data”)
|
||||
- Configure the node to read rows (default “read/get” behavior; keep defaults unless you need filtering)
|
||||
|
||||
4) **Add “Aggregate Rows” (Aggregate node)**
|
||||
- Connect **Read Google Sheets Data → Aggregate Rows**
|
||||
- Set **Aggregate** mode to **Aggregate All Item Data**
|
||||
- Set **Destination Field Name** to: `all_rows`
|
||||
|
||||
5) **Add “OpenAI Chat Model” (OpenAI Chat Model node)**
|
||||
- Choose model: **gpt-4.1-mini**
|
||||
- Add/attach OpenAI credentials (API key / OpenAI credential in n8n)
|
||||
- Leave options default unless you need temperature/max tokens
|
||||
|
||||
6) **Add “Generate Summary with AI” (AI Agent node)**
|
||||
- Connect **Aggregate Rows → Generate Summary with AI**
|
||||
- Connect **OpenAI Chat Model → Generate Summary with AI** using the **AI language model** connection type (not main)
|
||||
- Set Prompt type to **Define**
|
||||
- System message:
|
||||
- “You are a business analyst assistant. Write concise, professional weekly summary reports. Use bullet points and keep it under 300 words. Do not include greetings or sign-offs, just the report content.”
|
||||
- User prompt text (include expression for the data):
|
||||
- Provide instruction + embed: `{{ JSON.stringify($json.all_rows) }}`
|
||||
|
||||
7) **Add “Format Report” (Set node)**
|
||||
- Connect **Generate Summary with AI → Format Report**
|
||||
- Add fields:
|
||||
- `report_body` = expression referencing the agent output (as in the workflow): `{{ $json.output }}`
|
||||
- `report_date` = `{{ $now.format('yyyy-MM-dd') }}`
|
||||
- `subject_line` = `Weekly Report - {{ $now.format('yyyy-MM-dd') }}`
|
||||
- If your Agent node returns a different field name than `output`, adjust `report_body` accordingly.
|
||||
|
||||
8) **Add “Send Report Email” (Gmail node)**
|
||||
- Connect **Format Report → Send Report Email**
|
||||
- Configure Gmail OAuth2 credentials in n8n (connect the Gmail account authorized to send)
|
||||
- Set:
|
||||
- **To**: recipient email address
|
||||
- **Subject**: `{{ $json.subject_line }}`
|
||||
- **Message**: `{{ $json.report_body }}`
|
||||
|
||||
9) **Add “Log Report Sent” (Google Sheets node)**
|
||||
- Connect **Send Report Email → Log Report Sent**
|
||||
- Authenticate Google Sheets (can be same or different Google credential)
|
||||
- Set **Operation**: **Append**
|
||||
- Set **Document**: the Google Sheet ID used for logging
|
||||
- Set **Sheet name**: `Report Log`
|
||||
- Ensure the “Report Log” tab has headers/columns: **date**, **subject**, **sent_at**
|
||||
- Map columns:
|
||||
- `date` = `{{ $json.report_date }}`
|
||||
- `subject` = `{{ $json.subject_line }}`
|
||||
- `sent_at` = `{{ $now.toISO() }}`
|
||||
|
||||
10) **Test and activate**
|
||||
- Manually execute once (or temporarily change schedule to “every minute”) to validate:
|
||||
- Sheets read returns expected rows
|
||||
- Aggregate output looks correct
|
||||
- AI returns text in the expected field
|
||||
- Gmail sends successfully
|
||||
- Log row appends properly
|
||||
- Activate the workflow.
|
||||
|
||||
---
|
||||
|
||||
## 5. General Notes & Resources
|
||||
|
||||
| Note Content | Context or Link |
|
||||
|---|---|
|
||||
| This workflow runs automatically every Monday at 9am. It reads data from a Google Sheets spreadsheet, aggregates all the rows into a single payload, and sends it to OpenAI to generate a concise weekly summary report with key trends and highlights. The AI-generated report is formatted with a date-stamped subject line, emailed to your chosen recipient via Gmail, and logged to a separate spreadsheet so you have a record of every report sent. | From sticky note “How it works” |
|
||||
| Setup steps: 1) Adjust schedule 2) Connect Google Sheets + set spreadsheet ID & sheet name 3) Connect OpenAI credentials 4) Set recipient email 5) Connect Gmail 6) Create “Report Log” with columns: date, subject, sent_at 7) Set Report Log spreadsheet ID 8) Activate workflow | From sticky note “Setup steps” |
|
||||
| “Schedule & Read: Runs every Monday at 9am, reads all rows from the data spreadsheet, and aggregates them.” | From sticky note “Schedule & Read” |
|
||||
| “Summarize, Send & Log: AI generates a summary report, emails it via Gmail, and logs the send to a separate spreadsheet.” | From sticky note “Summarize, Send & Log” |
|
||||
Reference in New Issue
Block a user