WORKFLOW AUTOMATION & AI INTEGRATION

Project Overview

A client used one centralized Gmail account to receive invoices for several companies and suppliers. Preparing documents for quarterly tax filing in Germany required manually finding, downloading, identifying, renaming, and organizing every invoice—especially cumbersome when forwarded emails contained multiple attachments. I built an active n8n automation that processes every PDF attachment individually, uses Gemini to identify and extract data from invoices, and organizes valid documents across Google Drive, Gmail, and Google Sheets. Duplicate protection, confidence-based review, request-limit handling, and Google Chat error alerts make the process reliable and maintain human oversight when needed.

Responsibilities

• Designed and built the complete production workflow in n8n.
• Configured scheduled Gmail monitoring and attachment-level iteration so every PDF is processed, including multiple attachments within one email.
• Integrated Gemini for invoice classification and structured extraction of the billed company, supplier, invoice date, total amount, and currency.
• Created the company-code mapping, standardized filename logic, Google Drive destination rules, and duplicate-file protection.
• Built the hierarchical Company/Biller Gmail labeling process and configured completed messages to be marked as read.
• Added a Google Sheets monitoring log with document details, file links, processing status, and manual-review indicators.
• Implemented confidence-based review routing, controlled wait steps for Google service limits, and Google Chat failure notifications.
• Tested the workflow using multiple invoice formats, multi-attachment emails, duplicate documents, and PDFs that should not be classified as invoices.

Documented Workflow

1. Every two hours, n8n checks the centralized Gmail account for qualifying messages containing PDF attachments.
2. The workflow extracts every PDF and processes each attachment individually, including multiple files forwarded in a single email.
3. It extracts the document text while preserving the original PDF binary for later filing.
4. Gemini determines whether the document is an invoice and returns structured invoice data.
5. Non-invoice documents and uncertain results are excluded from automatic filing or routed for manual review.
6. For valid invoices, the workflow assigns the client-defined company code and generates a standardized filename and Google Drive destination.
7. Google Drive is checked for an existing file before a non-duplicate invoice is uploaded.
8. A monitoring row is added to Google Sheets with the processing time, company, supplier, invoice date, amount, currency, file link, status, and review requirement.
9. The required Company/Biller Gmail labels are created or reused, applied to the source message, and the message is marked as read.
10. Results below the required confidence level are flagged for verification of the company code, supplier name, and invoice amount.
11. A separate error workflow sends the failed execution, affected node, and error details to a Google Chat space for investigation.

Tools & Platforms

• n8n
• Gmail
• Google Drive
• Google Sheets
• Gemini
• Google Chat
• Google Workspace

Project Overview

This course project addressed a client-style requirement to automatically attach a Xero financial report to an Asana task after the task was completed. The requested report was not directly available through Xero's API, so the workflow could not simply download and forward the native report. I researched the available Xero API resources, selected the Account Transactions endpoint as the closest source, and built a Make.com scenario that reconstructs the required report as a CSV. The completed file is automatically attached to the corresponding Asana task.

Responsibilities

• Researched Xero's API documentation and available financial-data endpoints.
• Identified the Account Transactions endpoint as the closest available source for the requested report.
• Designed and configured the complete Make.com scenario.
• Connected completed Asana tasks to the report-generation process.
• Configured the Xero API request and mapped the returned transaction data.
• Used a router to coordinate data preparation and final file generation.
• Added an iterator to process individual transaction records.
• Used Google Sheets as a temporary staging area for normalized report rows.
• Configured the required delay to allow the transaction rows to finish populating.
• Retrieved the completed range and converted it into CSV-formatted text.
• Recreated the relevant structure and column order of the manually downloaded report.
• Uploaded the generated CSV file to the corresponding Asana task.
• Cleared the temporary Google Sheets range after the attachment was created.
• Tested the data retrieval, formatting, attachment, and cleanup sequence in Make.com.

Documented Workflow

1. The scenario watches Asana for tasks that have been marked complete.
2. Make.com calls the Xero Account Transactions endpoint to retrieve the closest available transaction-level data for the requested report.
3. A router coordinates the transaction-processing branch and the final report-generation branch.
4. The iterator separates the API response into individual transaction records.
5. Each normalized transaction is added to a temporary Google Sheets range using the required report structure.
6. A short delay allows all transaction rows to be written before file generation begins.
7. The scenario reads the populated range from Google Sheets.
8. A text aggregator converts the prepared rows into a CSV-formatted report.
9. The generated CSV is uploaded to the completed Asana task as an attachment.
10. The staging range in Google Sheets is cleared so it is ready for the next report run.

Tools & Platforms

• Make.com
• Asana
• Xero API
• Google Sheets
• Iterator
• Router
• Text Aggregator

Project Overview

This course project demonstrates how incoming form submissions can be automatically enriched, qualified, and routed according to their sales priority. I built a Zapier workflow that extracts the lead's email domain, retrieves additional company information from Apollo.io, and separates leads into high- and low-priority paths. High-priority leads receive immediate sales attention and personalized outreach, while lower-priority leads are forwarded to the sales team for manual follow-up.

Responsibilities

• Designed and configured the complete lead-qualification workflow in Zapier.
• Connected Youform as the lead-capture source.
• Used Formatter by Zapier to extract the company domain from each submitted email address.
• Configured a webhook request to enrich the lead through Apollo.io.
• Mapped the original form submission and enriched company data into the subsequent workflow steps.
• Created conditional High Priority and Low Priority paths.
• Configured Google Sheets to record qualified high-priority leads.
• Added real-time Slack notifications for the sales team.
• Used AI by Zapier to generate personalized outreach content from the available lead information.
• Connected Gmail to send the generated outreach to qualified leads.
• Routed lower-priority leads to the sales team for manual evaluation and follow-up.
• Tested the enrichment, routing, notification, and email steps during the available Zapier trial environment.

Documented Workflow

1. The automation begins when a prospect submits a lead form through Youform.
2. Formatter by Zapier isolates the domain from the submitted email address.
3. A webhook sends the domain to Apollo.io and retrieves additional company information for qualification.
4. Zapier Paths applies the configured qualification conditions and assigns the lead to either the High Priority or Low Priority path.
5. For a high-priority lead, the enriched information is recorded in Google Sheets so the sales team has a structured reference.
6. Slack sends an immediate alert containing the relevant lead information.
7. AI by Zapier uses the submitted and enriched information to prepare a customized outreach message.
8. Gmail sends the generated outreach email directly to the high-priority prospect.
9. For a low-priority lead, the information is sent to the sales team through Gmail for manual evaluation and follow-up rather than initiating automated outreach.

Tools & Platforms

• Zapier
• Youform
• Formatter by Zapier
• Webhooks by Zapier
• Apollo.io
• Google Sheets
• Slack
• AI by Zapier
• Gmail

Project Overview

This course project demonstrates how Asana can serve as a lightweight CRM while Zapier coordinates the operational work associated with each deal stage. I built a five-path automation that responds to Asana task updates and triggers the appropriate onboarding, follow-up, proposal, or customer communication process. The workflow keeps lead handling consistent while reducing the need to manually monitor every task and remember the next action.

Responsibilities

• Designed and configured the complete multi-path workflow in Zapier.
• Connected Asana as the CRM-style source of lead and deal-stage information.
• Created five conditional paths for the Ready to Start, No Response, Quoted, Approved, and Paid & Closed stages.
• Mapped Asana task information into Google Drive, Gmail, and AI-generated communications.
• Configured Google Drive folder creation for new client work.
• Added automatic Asana task creation for the next operational step.
• Built scheduled email follow-up sequences using Delay by Zapier.
• Added Gmail searches and filters to prevent unnecessary follow-up messages.
• Configured AI by Zapier to generate personalized customer communications.
• Created automated welcome and recommendation emails.
• Tested the routing conditions and actions during the available Zapier trial environment.

Documented Workflow

1. The automation begins whenever a tracked task is updated in Asana.
2. Zapier Paths checks the updated status and routes the task into one of five workflows.
3. For Ready to Start, a client folder is created in Google Drive, followed by a new Asana task for the next production activity.
4. For No Response, Gmail sends an initial lead follow-up. After a configured delay, the workflow searches for a relevant response and applies a filter before sending a second follow-up.
5. For Quoted, Gmail sends a quote follow-up. After a delay, the workflow checks for a response and sends a second quote reminder only when the filter conditions are met.
6. For Approved, the automation locates the client's Google Drive folder and required file, uses AI by Zapier to prepare personalized content, and sends a welcome email through Gmail.
7. For Paid & Closed, AI by Zapier generates a personalized recommendation message, which is then delivered to the customer through Gmail.

Tools & Platforms

• Zapier
• Asana
• Google Drive
• Gmail
• AI by Zapier