Solutions Showcase: part 4 of 7

The complete text of the site in 7 parts. Parts:

Part 1 has the contents list of every section and case study.

Case studies in full

Smart Amazon Product Tracker for Price Drop Alerts

To fetch product details from Amazon and alert you if the price drops, because who doesn’t love a good deal? 💸

Section: AI Agents

Project Overview

This project automates the process of monitoring Amazon product prices. Product links are stored in Airtable, where the workflow retrieves the records and sends product URLs to Apify for scraping. Once the scraping process is complete, the workflow retrieves the latest product information, updates the corresponding Airtable records, and checks whether the price has decreased.

When a price drop is detected, an email notification is sent to alert the user. The latest price is saved so it can be used for future comparisons.

01. The Challenge

Monitoring multiple Amazon products manually can be time-consuming and inconsistent. Prices may change between checks, making it difficult to identify savings opportunities quickly.

The main challenges included:

02. The Solution

An automated workflow connects Airtable, Apify, AI-assisted URL processing, and Gmail to handle product monitoring from beginning to end.

The workflow retrieves tracked products from Airtable, processes each product URL, launches an Apify scraping task, and waits for the results. It then updates the product record with the latest information and compares the new price against the previously stored price.

If the price has decreased, the workflow sends an email notification. If no price drop is detected, it updates the record without sending a price-drop alert.

03. How the Automation Works

Image: n8n working flow

Step 1 — Retrieve Products from Airtable

The workflow begins by retrieving product records from Airtable. These records contain the product information and links needed for price monitoring.

Airtable acts as the central tracking database, allowing the workflow to process multiple products without requiring each product to be entered manually into the scraping process.

Step 2 — Process Product URLs

The workflow loops through the retrieved product records and processes each product URL individually.

A Google Gemini-powered AI component helps format URLs when necessary, including handling quotation marks or other formatting issues that could interfere with the scraping request.

Step 3 — Launch the Apify Scraper

The processed product URL is submitted to Apify through an API request. Apify performs the scraping operation to retrieve available product details from Amazon.

This separates the scraping task from the rest of the workflow and allows the automation to process the results once the scraping operation has finished.

Step 4 — Wait for the Scraping Process

After launching the scraper, the workflow waits for the task to complete. It then requests the results from Apify.

This step ensures that the workflow retrieves the scraping output before attempting to update the product record or compare prices.

Step 5 — Update Product Information

The retrieved product details are written back to the corresponding Airtable record.

Keeping the product data in Airtable provides a centralized view of the tracked products and their latest recorded prices.

Step 6 — Check for a Price Drop

The workflow evaluates whether the latest product price is lower than the previously stored price.

This conditional step ensures that price-drop emails are sent only when the configured comparison identifies a decrease.

Step 7 — Send an Email Alert

When a price drop is detected, the workflow uses Gmail to send an email notification.

The alert informs the user that the tracked product has become cheaper, allowing them to revisit the product and decide whether to purchase it.

Step 8 — Maintain Updated Records

The workflow saves the latest price regardless of whether a price drop is detected. This keeps the stored information current and provides a reference point for subsequent monitoring runs.

04. Workflow

05. Technology Stack

TechnologyRole
AirtableStores product records and updated price information
ApifyScrapes available Amazon product details
Google GeminiAssists with product URL formatting
GmailSends email notifications when a price drop is detected
API RequestsConnect the workflow to the scraping service

06. Key Features

07. Business Benefits

Less Manual Work

Automates repetitive product-page checks, data collection, and record updates.

Faster Price-Drop Awareness

Email notifications help users discover price decreases without continually checking Amazon manually.

Better Data Organization

Airtable provides a central place to maintain tracked product records and current prices.

Consistent Monitoring

The workflow follows the same scraping, comparison, and update process for each product it processes.

08. Conclusion

The Smart Amazon Product Tracker demonstrates how API-based scraping, AI-assisted processing, conditional logic, and automated notifications can work together in a single workflow.

By connecting Airtable, Apify, Google Gemini, and Gmail, the solution automates product data collection, maintains updated price records, and notifies users when a tracked product becomes cheaper.

Page · Plain text

AI-Powered E-Commerce Product Price Monitoring

A scheduled workflow that tracks product prices, logs history to Google Sheets, and alerts you only on meaningful price drops.

Section: n8n Automations, AI Agents, AI Agents for Ecommerce

Project Overview

The AI-Powered E-Commerce Product Price Monitoring system automates product price tracking, historical data collection, and intelligent price-drop alerts. Built with n8n, Google Sheets, Slack, OpenAI, and Gemini, it regularly collects product prices, compares them with historical records, and identifies meaningful changes.

By automating repetitive price checks and notifying users only when relevant opportunities arise, the system helps users make informed purchasing decisions while saving time and avoiding unnecessary alerts.

Technologies: n8n · Google Sheets · Slack · OpenAI · Gemini

01. The Problem: Manual Price Checking Wastes Time

Before automation, tracking product prices across online stores was slow, repetitive, and error-prone.

Users had to manually:

This often resulted in missed price drops, limited-time deals, and a lack of historical visibility into price trends. Buying decisions were frequently based on incomplete information rather than reliable data.

Impact per user

Core challenge: Users want to buy at the right time without constantly monitoring product prices themselves.

02. Results After Automation

After implementation, the system automatically monitors prices, records historical data, and sends alerts when meaningful changes occur.

Key improvements

Time and effort savings

Before

After

Accuracy and opportunity gains

03. The Solution: AI-Powered Price Monitoring

The system acts as a virtual price tracker and analyst by combining scheduled data collection, historical tracking, AI analysis, and automated notifications.

It:

04. How the System Works

Phase 1: Scheduled Price Collection

The workflow runs automatically on a daily or custom schedule.

It visits configured product pages and extracts:

Incomplete or erroneous entries are skipped.

Phase 2: Price Comparison and History Logging

Phase 3: AI-Driven Analysis

OpenAI or Gemini evaluates the collected information to help identify:

Phase 4: Smart Notifications

Notifications are sent when configured conditions are met, such as a significant price drop or a better deal becoming available.

05. Business Value

06. Workflow Architecture

The automation follows this sequence:

Each component has a specific role:

07. How to Replicate the System

Step 1: Import the Workflow

  1. Open the attached E-Commerce Price Monitoring Automation.json file.
  2. Copy its complete JSON content.
  3. Open your n8n cloud or self-hosted instance.
  4. Create a new workflow and import or paste the JSON.
  5. Rename the workflow to E-Commerce Price Monitoring Automation.

Step 2: Configure Connections

Step 3: Test the Workflow

Run the workflow manually and verify that:

Step 4: Activate the Automation

Once testing is complete, activate the workflow and confirm the schedule is configured correctly.

The system will then monitor product prices, maintain historical records, analyze changes, and send alerts according to the configured rules.

08. Final Outcome

The AI-powered price monitoring workflow transforms repetitive, manual price checking into an automated and data-driven process.

Users can monitor multiple products, review historical prices, identify potential savings, and receive meaningful alerts without continuously checking online stores.

The result: Less manual work, better price visibility, and more informed purchasing decisions.

Page · Plain text

Ingredient Quoting Across 22 Supplier Portals

A Python pipeline (Prefect, Skyvern and GPT-4.1) that quotes all 22 approved suppliers for every ingredient reorder and delivers a ranked comparison in about four hours instead of 3.5 days.

Section: Python Automation

At a glance

Mid-size generic drug manufacturer, ~40 SKUs in production. Five weeks. Python 3.12, Prefect, Skyvern Cloud, OpenAI GPT-4.1, Postgres, S3, Slack. Went from quoting a fifth of their supplier base per order to quoting all of it.

The 6% nobody could see

Production planning flags a reorder for an active ingredient. A procurement officer opens a portal, searches the CAS number, pulls a price, downloads the spec sheet, pastes it into Excel. Then the next portal. Then the next.

Twenty-two suppliers were on the approved list. In practice the officer got through five or six before the purchase order had to go out, and the same five or six every time, the ones with the least annoying logins.

The waste wasn't the officer's afternoon. It was the sixteen vendors nobody had time to ask. When the full list was finally quoted, the cheapest compliant supplier turned out not to be one of the usual five surprisingly often.

Why this wasn't a scraping job

Three things ruled out the obvious approach, and a hand-written Playwright suite had already been abandoned for exactly these reasons.

Vendors don't agree on what anything is called. Metformin HCl USP, metformin hydrochloride BP/EP, or a bare product code, same molecule, three listings, no shared key except the CAS number buried somewhere on the page.

Quotes don't arrive in one shape. Some portals render a price table. Some hand back a PDF. Some won't tell you anything until you've submitted an RFQ form and waited for an email.

And it's a regulated purchase. Pharmacopoeia grade, GMP status, and whether the supplier holds a DMF or CEP are pass/fail gates, not preferences. A cheap quote from a vendor without the right filing isn't a cheap quote. It's not a quote at all. Every choice has to survive an audit two years later.

Following one RFQ through

Image: Multi-vendor ingredient quoting pipeline: RFQ intake (CAS number, grade, quantity), a Skyvern browser handles login, 2FA and CAPTCHA, prices come from an in-page table or from a downloaded and parsed PDF or emailed quote, GPT-4.1 normalises and validates them, and the ranked comparison goes to a Slack thread with an audit log.

A reorder request arrives as a CAS number, a grade, and a quantity. Before anything opens a browser, that gets resolved against an internal ingredient register, deterministically, on the CAS number, never on the product name.

Prefect then fans the request out to twenty-two isolated subflows, one per vendor. Each launches its own Skyvern cloud browser and logs in. Nine of these portals want 2FA and two sit behind CAPTCHAs; that's handled inside the login step rather than being a per-vendor engineering problem.

From there the paths diverge by portal. Where the price sits in the page, it's read straight off the page against a schema. Where it only exists as a PDF or an emailed quote sheet, the file is downloaded and parsed separately.

Everything then funnels back into a single normalisation pass, one model call, one strict schema, twenty-two vendors reduced to comparable rows: unit price, currency, unit of measure, MOQ, lead time, incoterm, grade, GMP status, DMF reference.

The ranked comparison lands in Slack about four hours later, with every row linking to the evidence behind it.

Four forks in the road

Skyvern rather than raw Playwright

The rejected option was writing and maintaining twenty-two scripts. That's what had already failed. At this many portals somebody is always redesigning something, and the maintenance load quietly exceeded the hours the automation was saving. Buying the resilience rather than building it was the entire reason the project was viable a second time.

1Password rather than a second vault

Skyvern can read credentials from several providers, and the client was already on 1Password. The alternative, standing up a dedicated secrets store, would have meant twenty-two sets of credentials living in two places, which is worse security dressed up as better engineering.

Selectors first, model second

Every stable element on a page gets a selector and a prompt. The selector runs, costs nothing, and usually works; the model only engages when it misses. Going model-first on everything would have been simpler to write and roughly ten times the running cost.

The model never decides what an ingredient is

It was tempting to let a model reconcile vendor product names. It isn't allowed to. Where a listing has no CAS field, the model may suggest a match, and that suggestion goes to a human. Misidentifying an ingredient in this industry isn't a data quality issue, and no accuracy figure would have made that trade acceptable.

When it breaks

Image: Route Memorization in practice: a memorized route replays deterministically. If the replay works the quote is captured with its evidence hash; if the route breaks it is regenerated by AI with a diff alert; if it keeps failing it goes to a review queue with the session recording attached.

Skyvern's Route Memorization compiles a successful run into a deterministic script and replays it. When a portal changes and the replay fails, it re-plans with AI, finds the new path, and regenerates the script. That mechanism does the heavy lifting here.

What it doesn't do is announce itself. A regenerated route is a silent event, and silent events don't belong in a regulated purchasing process.

So the wrapper watches for it. Each subflow records its action count and wall-clock duration; a memorised replay sits in a tight, predictable band, and an AI re-plan blows straight through it. That signal tags the run as a regeneration, diffs the new route against the old one, and drops a summary into Slack so someone can see what the portal changed before the next cycle.

If a regenerated route produces a quote that can't be evidence-checked, that vendor goes to a review queue with its session recording attached instead of into the comparison. About 4% of quotes land there, mostly from two portals that render prices as images.

The result is that failure is local. One portal breaking means twenty-one quotes still land and the RFQ is still useful.

Impact

BeforeAfter
Vendors quoted per RFQ5–622
Time to full comparison3.5 days~4 hours
Procurement hours per month~26~3

Page · Plain text

Monday.com Setup: Veterinary Specialist Partners

Monday.com board setup built for Veterinary Specialist Partners’ operations.

Section: Monday.com Setups

Project Overview

Veterinary Specialist Partners (VSP) operates specialized veterinary hospitals in Louisville and Evansville, providing care for pets with complex medical conditions.

Before the implementation, appointment scheduling, patient records, follow-up communications, and referrals relied heavily on manual processes. VSP partnered with our team to configure monday.com as a centralized operational workspace and automate key processes from appointment booking through patient follow-ups and referral management.

01. The Challenge

Manual Appointment Scheduling

Appointments were managed through phone calls and in-person scheduling, creating administrative bottlenecks and additional workload.

Fragmented Patient Records

Patient histories, treatment records, and supporting documents were spread across different platforms, making information harder to retrieve and manage.

Inconsistent Follow-Up Communications

Manual follow-up scheduling and patient reminders were prone to delays and errors.

Referral Coordination

Managing referrals between specialists required better tracking and coordination to ensure patients were not overlooked.

02. Project Objectives

The implementation focused on three primary objectives:

03. Solutions Implemented

A. Automated Appointment Management

New appointments booked through the configured process are captured in a dedicated monday.com appointments board.

B. Calendly Integration

Calendly is integrated with monday.com to synchronize appointment bookings and cancellations.

C. Email and SMS Reminders

Automated reminders notify patients about upcoming appointments.

D. Centralized Patient Records

monday.com serves as the central workspace for organizing patient-related information.

E. Follow-Up Task Automation

Follow-up tasks are created and linked to specific appointments.

F. Referral Management

A dedicated monday.com board manages referrals to and from other specialists.

04. System Workflow

This workflow brings appointment scheduling, patient information, communication, and ongoing care coordination into one operational process.

05. Outcomes and Business Value

Improved Operational Efficiency

Automating appointment handling reduces repetitive administrative work and helps minimize errors associated with manual scheduling.

Centralized Data Management

Organizing patient histories, treatment records, and appointment information in monday.com makes records easier to access and supports more informed operational decisions.

More Reliable Patient Communication

Automated email and SMS reminders reduce dependence on manual messaging and help patients stay informed about upcoming appointments.

Better Follow-Up Coordination

Linked follow-up tasks provide a structured way to track post-appointment activities and outstanding work.

Improved Referral Management

A dedicated referral board and associated workflows help organize referrals and coordinate activities between specialists.

The original case study describes these qualitative outcomes but does not provide numerical measurements for time saved, no-show reduction, or error reduction.

06. Technical Highlights

monday.com Customization

Integrations

The exact configuration of each external integration depends on the connected services and deployed automation rules.

07. Future Improvements

Ongoing Optimization

Continue refining boards and automation rules based on team feedback and changing operational requirements.

Expanded Automation

Identify additional manual processes across hospital operations that could benefit from automation.

Workflow Refinement

Review appointment, follow-up, and referral workflows regularly to ensure they remain aligned with the hospital's operational needs.

08. Final Outcome

The monday.com setup provides Veterinary Specialist Partners with a more structured way to manage appointments, patient information, reminders, follow-ups, and referrals.

By centralizing operational records and automating repetitive coordination tasks, the solution helps reduce manual administrative effort and supports more consistent communication across the veterinary care process.

Page · Plain text

AI-Powered Social Media Content Creation

An automation that generates LinkedIn captions and images, sends them for approval, and publishes only approved posts.

Section: n8n Automations, AI Agents, AI Agents for Ecommerce

Project Overview

The AI-Powered Social Media Content Creation system automates the process of generating, reviewing, and publishing LinkedIn content. Built with n8n, LinkedIn, Google Sheets, and Google Drive, it uses AI to create captions and matching post images while maintaining a centralized record of content and approval status.

By introducing an approval step before publishing, the system helps teams maintain consistent brand messaging, reduce repetitive work, and manage their social media content through a single workflow.

Technologies: n8n · LinkedIn · Google Sheets · Google Drive · AI Content Generation

01. The Problem: Social Media Management Is Inconsistent

Before automation, creating and publishing social media content required multiple tools and manual steps.

Teams had to:

Impact per team

Core challenge: Teams needed a reliable system to create, review, approve, and publish content at scale without unnecessary operational complexity.

02. Results After Automation

After implementation, content creation and publishing became a structured, automated process.

Key improvements

Time and effort savings

Before

After

Monthly impact

Consistency and tracking improvements

03. The Solution: AI-Powered Social Media Automation

The system acts as a virtual social media assistant that manages content from creation to publication.

It:

04. How the System Works

Phase 1: AI Caption Generation

The workflow starts when a new row is added to Google Sheets.

AI generates captions based on:

The captions are formatted for LinkedIn and saved in Google Sheets for review and approval.

Phase 2: Image and Post Creation

The generated caption is used to create a matching AI-generated image.

Phase 3: Approval and Publishing

Google Sheets is updated with:

This provides a centralized record of the content lifecycle and helps prevent missed or duplicate publishing.

05. Business Value

06. Workflow Architecture

Each component has a defined role:

07. How to Replicate the System

Step 1: Import the Workflow

  1. Open the attached Social Media Automation Workflow.json file.
  2. Copy the complete JSON content.
  3. Open your n8n cloud or self-hosted instance.
  4. Create a new workflow and paste or import the JSON.
  5. Save and rename it to Social Media Content Automation.

Step 2: Configure Connections

Step 3: Test the Workflow

Run the workflow manually and verify that:

Step 4: Activate the Automation

Once testing is complete, activate the workflow and verify its trigger settings.

Content generation, approval, and publishing can then run automatically with minimal manual involvement.

08. Final Outcome

The AI-powered workflow transforms social media management into a consistent and trackable process.

The result: A more efficient, approval-controlled social media workflow that allows teams to scale content production without proportionally increasing manual effort.

Page · Plain text

AI-Powered Whatsapp message Assistant

An AI-powered WhatsApp assistant that automates customer support through text, voice, and image understanding, delivering instant, knowledge-based responses 24/7.

Section: n8n Automations, AI Agents, AI Agents for WhatsApp, AI Agents for Ecommerce

01.Project Overview

The AI-Powered WhatsApp Message Assistant is an intelligent customer support solution built with n8n, OpenAI, WhatsApp, and Supabase. It automates customer conversations by understanding text, voice messages, and images, then generating relevant responses based on business knowledge.

Using AI-powered processing and semantic search, the assistant delivers consistent, context-aware support around the clock, reducing repetitive manual work and allowing human agents to focus on more complex customer inquiries.

02. The Problem: WhatsApp Customer Support Didn’t Scale

Before automation, customer support over WhatsApp was entirely manual and increasingly difficult to manage.

Staff had to respond to every message individually, including repetitive questions about products, prices, availability, and store policies. Voice messages required manual listening and interpretation, while image-based questions, such as product photos, needed human review. Responses varied depending on the staff member, and support was unavailable outside business hours.

For a small business receiving 80–120 WhatsApp messages per day, this resulted in:

Core Challenge: WhatsApp was the main customer communication channel, but human-only support could not keep up with demand without increasing headcount.


03. Results After Automation

After deploying the WhatsApp AI Business Assistant:

Time and Workload Impact

Before Automation

After Automation

Estimated Time Savings

This allowed the business to scale WhatsApp support without hiring additional staff.


04. The Solution: An AI Assistant Inside WhatsApp

The system acts as a virtual support agent directly within WhatsApp.

Customers receive fast, helpful support without needing to change how they communicate with the business.


05. How the System Works

  1. Customer Message: Customers contact the business through WhatsApp.
  2. Message Understanding: AI identifies the message type and intent.
  3. Knowledge Retrieval: Relevant business information is retrieved using semantic search.
  4. Response Generation: AI generates a context-aware answer based on the available information.
  5. WhatsApp Reply: The response is delivered automatically to the customer.

The workflow processes messages automatically within seconds.


06. Business Value


07. Reliability and Safety

These measures help maintain reliable, consistent, and responsible customer support.


08. How to Replicate the System

The solution uses the WhatsApp Cloud API, n8n, OpenAI, and Supabase Vector Store.

Step 1: Set Up WhatsApp Cloud API

Configure the WhatsApp Business account, phone number ID, and webhook events for incoming messages.

Ensure the integration supports receiving text, voice, and image messages and sending text or audio replies.

Step 2: Create the WhatsApp Webhook in n8n

This connects incoming WhatsApp messages to the automation workflow.

Step 3: Import the n8n Workflow

Create a new workflow in n8n and import the provided file:

AI-Powered WhatsApp Assistant.json

Step 4: Connect AI and Data Services

OpenAI

Supabase Vector Store

This enables retrieval-augmented generation (RAG) for business-grounded responses.

Step 5: Configure Message Routing

Open the Switch node and verify the routing rules:

This separates message-processing paths according to the input type.

Step 6: Test the Workflow

Send test messages through WhatsApp, including:

Verify correct intent detection, relevant answers from business knowledge, successful WhatsApp replies, and voice responses when triggered.

Step 7: Activate the Workflow

Switch the n8n workflow to Active after testing and validating the integrations.

The assistant can then respond to customer messages automatically through WhatsApp.


09. Final Outcome

With the WhatsApp AI Business Assistant deployed:

The system transformed WhatsApp from a manual support burden into an always-on, AI-powered customer assistant.

Page · Plain text

ClickUp Structure: Boots & Bags (Ecommerce Dropshipping)

Detailed ClickUp workspace structure for an ecommerce dropshipping store’s operations.

Section: ClickUp Setups

ClickUp Structure: Boots & Bags (Ecommerce Dropshipping)

A structured ClickUp workspace designed to manage ecommerce dropshipping operations, including order fulfillment, customer service, product management, marketing, website maintenance, and branding.

Category: ClickUp Setups Technology: ClickUp Integrations: Shopify, Facebook, Google Ads, WhatsApp

01. Project Overview

The Boots & Bags E-Commerce Operations workspace organizes the store’s major business processes within ClickUp. Dedicated folders, lists, custom fields, templates, and automation rules help teams manage orders, resolve customer issues, coordinate product listings, monitor marketing activities, and maintain website quality.

02. The Business Challenge

Ecommerce dropshipping operations involve several connected activities that need to be coordinated efficiently.

Without a structured workspace, these activities can become difficult to track and coordinate across teams.

03. The Solution

A dedicated ClickUp workspace was organized under the main folder Boots & Bags E-Commerce Operations, with five operational areas:

  1. Fulfillment
  2. Customer Service
  3. Collection and Product Pages
  4. Marketing
  5. Website and Branding

Each area contains purpose-specific lists, relevant custom fields, and automation rules to support task tracking and follow-up.

04. Workspace Structure

A. Fulfillment

Manages order processing, stock shortages, disputes, and reshipment activities.

Orders Management

Out of Stock

Disputed Orders

Order Reshipment

Templates

B. Customer Service

Organizes customer requests and coordinates approved reshipments with the fulfillment team.

Reshipment Requests

Template

C. Collection and Product Pages

Coordinates product discovery, approval, and preparation for Shopify publication.

Product Research

Product Upload

Templates

D. Marketing

Organizes paid advertising and direct marketing campaigns across multiple channels.

Facebook Marketing

Google Ads and Bing Marketing

Email, SMS, WhatsApp, and Push Notifications

E. Website and Branding

Supports website health, brand consistency, and ongoing user experience improvements.

Website Health Checks

Custom fields:

Automation rules:

Branding Consistency Reviews

Custom fields:

Automation rules:

User Experience Optimization

Custom fields:

Automation rules:

05. End-to-End Workflow

This workflow represents the overall operational structure. Individual automation rules handle specific events, such as low inventory, approved reshipments, urgent website issues, and follow-up reviews.

06. Key ClickUp Capabilities

07. Expected Business Value

The workspace is designed to support:

08. Conclusion

The Boots & Bags ClickUp workspace provides a structured foundation for managing ecommerce dropshipping operations. By bringing fulfillment, customer service, product management, marketing, and website maintenance into one organized environment, it helps teams coordinate work, track responsibilities, and apply repeatable processes across the business.

Page · Plain text

Event-Driven Quote Triage for a Freight Brokerage

An event-driven Python service (FastAPI, Celery, Playwright and Claude) that answers load quote requests from a shared inbox with three ranked carriers. Median first reply: 48 seconds, down from 7 minutes 40.

Section: Python Automation

Overview

This case study focuses on an event-driven Python automation system that processes freight quote requests from a shared Gmail inbox. It checks carrier availability through browser automation, compares rates using Airtable data, and uses Claude to rank the top three carriers before sending a response through Gmail and Slack.

The system uses FastAPI, Celery, Redis, PostgreSQL, Browserbase, Playwright, and Claude. It replaces slow, unreliable polling with Gmail push notifications, uses database-based deduplication to prevent duplicate replies, and includes retries, recovery checks, and Slack alerts for failures.

Key results reported in the case study:

The problem

Load quote requests landed in a shared quotes@ inbox, around 40 a day. A dispatcher would read the request, log into the carrier portal (no public API for what they needed), check which trucks were available on that lane, cross-reference an internal rate sheet, and reply with three carriers and a price. Six to nine minutes each, every time.

Their first automation attempt was a Python script on a five-minute cron driving a hosted agent platform. It worked about 70% of the time, cost roughly $0.42 per request, and failed silently. They usually found out from a customer, not a log.

Three root causes, and all three shaped the rebuild:

The architecture

Everything runs as a small Python service on a single Fly.io machine, a FastAPI app that receives Gmail push notifications, and a Celery worker pool that does the actual work.

Image: Event-driven quote triage pipeline: a FastAPI webhook verifies and acknowledges the push, the request is de-duplicated and queued (Postgres key, Redis queue), Browserbase and Playwright check portal availability while an Airtable rate sheet supplies lanes, margins and exclusions, Claude ranks the top 3 carriers, and a threaded reply goes out on Gmail and Slack.

Key engineering decisions

Real-time trigger, not a cron. Gmail's users.watch publishes to a Google Pub/Sub topic, which pushes to a FastAPI endpoint. The handler does three things and nothing else: verify the JWT on the push request, write the message ID to Postgres, and return 204. Pub/Sub redelivers anything not acked within 10 seconds, so all real work is handed to Celery and the endpoint stays under 50 ms.

@app.post("/gmail/push")
async def gmail_push(req: Request, claims=Depends(verify_pubsub_jwt)):
    envelope = await req.json()
    history_id = decode_message(envelope["message"])["historyId"]
    for msg in gmail.list_since(history_id):
        if requests_repo.claim(msg.thread_id):   # INSERT ... ON CONFLICT DO NOTHING
            process_quote.delay(msg.thread_id)
    return Response(status_code=204)

Idempotency lives in the database, not in memory. claim() is an INSERT ... ON CONFLICT DO NOTHING on the Gmail thread ID. If Pub/Sub redelivers, or the safety net and a live webhook race each other, exactly one worker wins the row. This is the piece that eliminated duplicate replies outright.

Cached navigation instead of per-run LLM planning. Playwright drives a Browserbase session with stored credentials and a persistent context. The first time the portal layout is seen, an LLM produces a navigation plan, a sequence of selectors and actions, which is hashed against the page structure and cached in Postgres. Every subsequent run replays the cached plan with plain Playwright calls: no LLM, no tokens, sub-second steps. Only when a replay step fails does the code fall back to re-planning, write the new plan, and post a Slack notice that the portal changed. That's the same idea as Skyvern's Route Memorization, built explicitly so the failure path is visible rather than silent.

Prompt caching on the ranking call. The system prompt, the scoring rubric, and the serialised carrier roster are static across requests and sit in a cached block; only the parsed request and the fresh availability data go in the user turn. Roughly 11k stable input tokens moved to cache reads.

system=[{
    "type": "text",
    "text": RUBRIC + roster_block,
    "cache_control": {"type": "ephemeral"},
}]

Concurrency capped deliberately. The Celery pool runs three workers. The portal rate-limits aggressively and had previously locked the shared account twice, so throughput was traded for account safety.

State machine and the safety net

Every request row moves through an explicit status column. The hourly sweep is the belt-and-braces layer for the rare case where a Pub/Sub push is lost or a worker dies mid-run.

Image: Request states and recovery: a request is processing (row claimed, timestamp set), then done (reply sent, thread id) on success or failed (alert posted to Slack) on error. An hourly sweep re-runs rows stuck for more than 15 minutes.

The sweep is a Celery Beat task, about 30 lines: select rows in processing with claimed_at < now() - interval '15 minutes', reset the attempt counter, re-enqueue. After three attempts it moves to failed and posts a Slack alert with the thread link and the last exception, so a dispatcher can handle it manually and an engineer can see exactly where it broke.

Errors are handled in three tiers rather than one generic retry: transient network and portal timeouts retry with exponential backoff inside the task; a failed selector replay triggers re-planning and a "portal layout changed" notice; anything else fails the row loudly.

Results

BeforeAfter
Time to first reply (median)7 min 40 s48 s
Cost per request$0.42$0.06
Success rate (30-day)71%98.6%
Duplicate replies / week4–60
Silent failuresroutinenone, every failure alerts

The cost drop came from two places roughly equally: cached navigation removed the per-run planning tokens, and prompt caching cut input cost on the ranking call.

Delivered

Dockerised FastAPI + Celery service with a one-command deploy; Postgres schema and migrations; the Browserbase automation module with cached-plan replay and fallback; the Claude ranking module with the rubric prompt; Slack alerting; a runbook covering the four common failure modes and how to re-run a request by thread ID; and a 45-minute handoff call testing a live request end to end.

Page · Plain text

One click after the consultation creates a ready-to-work matter

CASE STUDY · BREWSTER LAW FIRM

Section: n8n Automations

Over View

This project automates Brewster Law Firm’s client intake process using Clio and n8n. When an attorney marks a consultation as Ready to Engage, the system automatically creates the matter, assigns the attorney, and sends the client a branded questionnaire with a secure link.

~1 min31 click10
from Ready to Engage topractice-area groups,by the attorney;connected workflows with
a created mattereach with its own starting stageeverything after it is automaticautomatic failure alerts

Once submitted, responses are saved directly in Clio, matter stages are updated, and relevant notifications are sent automatically. The solution reduces manual work, streamlines client onboarding, and keeps matter records organized.

THE PLAN OF ACTION

We started from a written plan, then built it part by part

Brewster Law Firm's clients arrive through Google and AI search, the website form, the booking page and the phone. Our Plan of Action found that what happened after a lead arrived depended on two people remembering every step. It set out a client journey in five parts, from inquiry to closing, with Clio as the single place where every lead, client and matter lives. Below is each problem the plan found and what we built for it.

Problem in the planWhat we builtStatus
Form submissions (Part 1)An instant, branded thank-you with the next step andDelivered
Leads reach Clio Grow, but the leadbooking link, then timed nudges that stop the moment
gets no thank-you email or nextthe lead books.
step.
Consultation bookings (Part 2)Confirmation and follow-up emails driven from ClioDelivered
No automatic confirmation emailGrow, alongside Clio's own appointment reminders.
after a booking.
After the consultation (Parts 2–3)One click sets the outcome to Ready to Engage. TheDelivered
Intake depends on someonematter is created in the right practice area and first
remembering each step: convert thestage, the attorney is assigned and the questionnaire
lead, create the matter, pick thegoes out.
stage, send the questionnaire.
Client questionnaires (Part 3)A branded questionnaire on the firm's own address, withDelivered
Intake forms are collected by hand.a one-time private link per matter. Answers are saved as a note on the matter.
Client updates (Part 4)A plain-language email on every stage change, a "stillDelivered
Some clients have to ask for a statusworking on it" note when a matter sits too long, and
update.reminders when the client owes something.
Info inbox (Part 1)Reply tracking pauses automated reminders the momentDelivered /
Replies are manual, usually the nexta client writes back. Reply drafting is the next phase.next phase
morning.
Problem in the planWhat we builtStatus
Ownership and control (Part 0)Clio stays the single place for leads and matters.Delivered
Tools are only partly set up andAutomations run on a firm-owned, self-hosted n8n
spread across chat and email.server, with a daily health report and instant failure alerts.
Agreements and payments (PartKept manual in Clio by the firm's own choice. Billing andBy design
3)agreements are deliberately not automated.
Approvals travel by chat;
agreements are hard to track.
Phone, documents, closing (PartsVoice intake, OneDrive sync and closing and reviewNext phase
1, 4, 5)emails are scoped in the plan for the next phase.
Smith.ai re-entry, OneDrive folders,
and no closing or review request.

One click by the attorney after the consultation. Everything after it happens on its own.

THE JOURNEY AT A GLANCE

The plan, drawn out

The five-part client journey. Green is built; dashed grey is planned for the next phase.

From consultation to questionnaire. One click, then the checks, matter, email and stage update happen on their own.

How the pieces connect. Clio stays the system of record; n8n runs on a firm-owned server.

HOW IT WORKS

From consultation to questionnaire

  1. The outcome is set to Ready to Engage The only manual step. It happens in Clio Grow, where the team already works.
  1. The matter is created automatically Formation matters start at "Questionnaire Sent", compliance matters at "Information Gathering", contract matters at "Documents Received". Before it creates anything, the system confirms the client has an email address, matches exactly one contact, and has no open matter. If anything looks unusual, it hands the case to a person instead of guessing.
  1. The client receives their questionnaire The email carries a private link tied to that one matter.
  1. The client submits it The answers are saved as a note on the matter in Clio, the link closes, and the matter moves to its next stage.
  1. Everyone is kept informed The client receives the update for the new stage. The firm receives a short "questionnaire received" notice.

WHAT IT LOOKS LIKE

Inside Clio and n8n

The matter-creation workflow. It reads leads from Clio Grow, applies the safety checks, creates the matter, and sends the two emails.

The client questionnaire. Hosted on the firm's own address, in Brewster's branding. Only the description of the matter is required.

The matter after the client submitted the questionnaire. It started in the compliance group's first stage and moved to Assessment in Progress on its own. Client details are blurred.

The answers land on the matter. The note is saved where the attorney will look for it.

WHAT THE CLIENT RECEIVES

Emails that read like the firm wrote them

Every email uses the firm's branding and Elissa Brewster Langston's own signature. These are real messages captured from the firm's mailbox during the verification run.

The questionnaire email. Sent the moment the matter is created, with a private link for that one matter.

The signature and confidentiality notice match the attorney's own email signature.

The stage update. When the questionnaire was submitted and the matter moved, the client received this note on its own. The [TEST] tag marks messages sent during verification.

BUILT TO BE TRUSTED

Guardrails a law firm needs

The firm's own voiceEvery client email is written in the firm's tone and carries the attorney's own
Per-matter controlsignature and confidentiality notice. A single Clio field turns automatic emails on or off for any matter.
Private by designEach questionnaire link works once, for one matter. A reused or invalid link is refused and recorded.
Your data stays yoursThe workflows run on a dedicated self-hosted server, not a shared automation service. Client records stay in Clio.

THE RESULT

The whole journey runs without hand-offs

Run end to end in Brewster's own Clio account, across all three practice-area groups, the system did what the team used to do by hand:

Matter createdIn about a minute, in the correct practice area and first stage, with the attorney
Questionnaire sentassigned. A private link arrived in the client's inbox at the same moment, in the firm's branding.
Stage movedFormation, compliance and contract matters each advanced to their next stage when the form came back.
Answers on the matterSaved as a note in Clio, with the team notified, and no one retyping anything.

RUN YOUR INTAKE THE SAME WAY If your firm uses Clio and still triggers intake by hand, we can map your stages, practice areas and client emails into a workflow like this one. Talk to Growlyze about a free intake review.

Page · Plain text

Automated Biological Age Engine

A health and wellbeing company sells a biological ageing assessment. It does not give one “body age”. It shows how each of 16 body systems, from heart and brain to sleep, digestion and skin, seems to be ageing…

Section: n8n Automations, Python Automation

Overview

A health and wellbeing company sells a biological ageing assessment. It does not give one “body age”. It shows how each of 16 body systems, from heart and brain to sleep, digestion and skin, seems to be ageing compared with the person’s real age, and what is driving each one.

Each client answers a 150-question form in JotForm and takes a short camera face scan in Upvio. The company had a locked scoring method written up as a specification, but nothing to run it: no way to join each form to its scan, turn the answers into system ages, or produce the personal report it promises.

Now an engine built by Growlyze, with three n8n workflows around it, matches each form to its scan, scores both in code exactly as the specification says, has AI write the report around those numbers, checks every part, and emails the client a private link to their PDF.

The Problem

Clients pay for a detailed, personal picture of their health. A wrong number or a careless sentence about their body would cost the company their trust.

Two tools, no shared ID. The form and the scan arrive separately, at different times and in either order, and the form carries no assessment number. Someone would have to pair every form with its scan by hand.

The maths has to be exact. The 150 answers feed 16 body systems, 58 clusters, 28 interaction rules and several indices, with weights that shift when a layer is missing. The same answers must always give the same ages, which an AI model cannot guarantee on its own.

The scan has to be handled honestly. A camera reading from one sitting is an estimate. Impossible readings must be left out, not scored, and the report must never name a medical condition.

The words have to be safe, and every client needs a report. The text must be in plain British English, must never change or invent a number, and must never make a diagnosis. A weak AI draft cannot be allowed to leave a paying client with nothing.

How It Works

  1. The form comes in. An n8n workflow maps the JotForm fields to the engine’s format and sends them on. Each answer must match one of the question’s exact options; an unknown answer stops that assessment and raises an alert instead of guessing. Repeats of the same submission are ignored.
  2. The scan comes in. A second n8n workflow keeps only completed scans from this assessment’s Upvio link, looks up the client’s email through Upvio’s patient API, and sends the scan to the engine.
  3. Form and scan are matched by email. Whichever arrives first is stored and waits for the other. When both exist, the engine links them under one assessment ID and queues the report. A scan is never reused for a second report.
  4. Code does the scoring. The answers and the usable scan readings (blood pressure, resting heart rate, breathing rate and waist-to-height ratio) are scored by the locked method into 16 system ages, burdens and directions. Missing answers are left out, never counted against the client, and readings outside what is physically possible are listed and excluded.
  5. AI writes the words. The AI gets a pseudonymised summary with no name or contact details and writes the report in parts. It never calculates: the tables and charts are filled by code, and any figure the AI mentions must be one the code already worked out.
  6. Every part is checked. Checks look for structure, changed or invented numbers, diagnosis claims and hard-to-read sentences. A failing part goes back for repair; minor issues that remain become internal warnings, and any unsafe sentence still there is removed by code, with an alert.
  7. The PDF is built. The engine adds the client’s first name, builds a branded report of around 45 pages, stores it and creates a private, unguessable link.
  8. The client is emailed. The engine tells n8n the report is ready, and a third workflow emails the client their link through Gmail, checking first that the event has not already been handled.

All of this runs as its own service, with its own database and four workers, beside the company’s live ADHD assessment without touching it. If the AI service fails, the job is retried and finished parts are kept. The scoring method and the scan mapping are fixed, fingerprinted files, so any change is a new, recorded version rather than a quiet edit.

Results & Metrics

A form and a scan now find each other by email and become a scored, written, checked report without anyone stepping in. The numbers come only from code, so the same answers always give the same ages, and every accepted assessment ends in a report.

MetricObserved Output
Test reports with the live AI model19
Processing time per report (current version)3 to 3½ minutes (5 of 5 runs)
Report length16 body systems, about 45 pages
AI drafts checked before publishing293
Reports sent with banned or unsafe wording0
Deliberate wrong answer and repeated webhookStopped with an alert; ignored, no second report
Automated tests887 passing (100%)

💡

Metrics come from development and test runs on made-up participants, timed from scored answers to stored PDF. Runs that were restarted after a fix are left out of the timing. Figures from live client traffic may vary.

Image: Workflow: the form intake in n8n.

Image: Workflow: the scan intake in n8n, with the email lookup in Upvio.

Image: Output: the job log of one test assessment with the live AI model, from form to notification in 3½ minutes. IDs shortened.

Image: Workflow: the report email in n8n, sent once per client.

Image: End result: two pages of the PDF report for a made-up test participant. Company name blurred.

Tech Stack

TechnologyRole in the solution
JotFormThe 150-question assessment form clients fill in
UpvioThe camera face scan, and the patient lookup that supplies the email
n8nThree workflows: form intake, scan intake, and the report email
Python engine (FastAPI)Matching, scoring, AI calls, checks, the job queue and the private report links
MySQL / MariaDBIts own database for assessments, scans, scores and the job queue
OpenAIWrites the report text around numbers set by code
WeasyPrintTurns the finished report into a branded PDF
GmailSends the client their report link

The engine runs on the company’s own server, next to its existing ADHD assessment but fully separate from it. Clients use the same kind of form and scan they already know.

Where This Pattern Applies

This isn’t really about ageing.

Worked example: a corporate wellness provider. Employees answer a lifestyle questionnaire and wear a fitness tracker for a week. The two are matched by email, a fixed scoring model turns them into risk areas in code, AI writes each person’s summary around those scores, the text is checked for medical claims, and a private report goes to each employee.

The shape: separate inputs are matched to one person, a fixed method scores them in code, AI writes within strict limits, and checks decide what gets sent. What changes from case to case is the questions, the scoring method, the second data source and the list of claims that must never appear.

Page · Plain text

Automated Health Assessment Engine

A health and wellbeing company sells an ADHD profile assessment. It is not a diagnosis and not a single “mild to severe” score. It shows how a person’s attention, drive, ideas, order, recovery and body work together,…

Section: n8n Automations, Python Automation

Overview

A health and wellbeing company sells an ADHD profile assessment. It is not a diagnosis and not a single “mild to severe” score. It shows how a person’s attention, drive, ideas, order, recovery and body work together, and it gives strengths and difficulties equal weight.

Each client completes a 72-question form in JotForm and a short camera body scan in Upvio. Before, the answers and the scan sat in those two tools with nothing joining them and nothing turning them into the personal report the company promises.

Now an assessment engine built by Growlyze, with three n8n workflows around it, matches each form to its scan, scores the answers in code, has AI write the report within strict limits, checks every part before publishing, and emails the client a private link to their PDF.

The Problem

People taking an ADHD assessment are often anxious about what it will say. They have paid for something personal, and a careless report can hurt them.

Two tools, no shared ID. The form and the scan arrive separately, at different times and in either order. Upvio’s scan notification does not even include the client’s email, so someone would have to pair every form with its scan by hand.

The scoring has to be exact. The 72 answers feed 11 dimensions, scoring bands, nine operating profiles, a seven-day experiment and product-suitability rules. The same answers must always give the same result, which an AI model cannot guarantee on its own.

The words have to be safe. The report must use the client’s own words in British English, but never claim a diagnosis or a cause, never mention brain chemicals or medication changes, and never show raw scores or question numbers. A short scan shows how the body was working at that moment; it must not be read as “how much ADHD” someone has.

Nothing can be lost or mixed up. Webhooks get sent twice, the AI service can be slow or down, and many clients can finish at once. Every accepted assessment still has to end in one report, sent once, to the right person.

How It Works

  1. The form comes in. When a client submits the form, an n8n workflow maps the JotForm fields to the engine’s format, checks it is complete and sends it to the engine. Repeats of the same submission are ignored, and unusable ones (missing answers, no consent) are rejected.
  2. The scan comes in. A second n8n workflow keeps only completed remote scans, looks up the client’s email through Upvio’s patient API, and sends the scan to the engine.
  3. Form and scan are matched by email. The engine stores whichever arrives first and waits for the other, with no time limit. When both exist, it links them and queues the report. If someone submits twice, the newest wins, and a scan is never reused for a second report.
  4. Code does the scoring. The answers are scored against the company’s rules into dimensions, bands, profiles, an experiment and suitability flags. Each scan reading (heart rate variability, recovery, stress, breathing, heart rate, blood pressure and a wellness score) becomes a plain finding and is compared with the questionnaire.
  5. AI writes the words. The AI gets a pseudonymised summary, with no name or contact details, and is told exactly which findings and which of the client’s own reflections it may use. It writes the questionnaire sections first and the scan section last.
  6. Every part is checked. The checks look for correct structure, evidence that points to real findings, use of the client’s words, and anything banned: question numbers, raw values, diagnosis or cause claims, medication advice, hype or US spelling. A failing part gets up to two targeted repairs; if it still fails, safe text built from the scores is used, so every client gets a report.
  7. The PDF is built. The engine adds the client’s name, builds a branded PDF, stores it and creates a private, unguessable link.
  8. The client is emailed. The engine tells n8n the report is ready, and a third workflow emails the client their link through Gmail, checking first that they have not already been sent it.

All of this runs in a database-backed job queue with several workers. If the AI service fails, the job is retried, and parts that already passed are kept, not written again. The questions, scoring rules and scan ranges live in one versioned configuration file, so a method change, or a whole new assessment, is a configuration edit rather than new code.

Results & Metrics

The team no longer pairs forms and scans by hand. Each client’s form and scan find each other by email, and a report is scored, written, checked and emailed without anyone stepping in. Nothing reaches a client until it passes the checks, and a temporary AI failure no longer means a missing report.

MetricObserved Output
Test runs with the live AI model14
Typical processing time per reportAbout 1 minute (median)
Processing time range39 seconds to just over 2 minutes (13 of 14 runs)
AI drafts checked before publishing66
Reports sent with banned or unsafe wording0
Load test1,000 submissions, none lost, duplicated or mixed up
Automated tests110 passing (100%)

💡

Metrics come from development and test runs, measured from validated input to stored PDF. The one run outside the range waited on automatic retries. Figures from live client traffic may vary.

Image: Workflow: the form intake in n8n.

Image: Workflow: the scan intake in n8n, with the email lookup in Upvio.

Image: Output: one engine run for a test participant, from matched form and scan to “ready”.

Image: Workflow: the report email in n8n, sent once per client.

Image: End result: two pages of the PDF report for a fictional test participant. Logo and company name blurred.

Tech Stack

TechnologyRole in the solution
JotFormThe 72-question assessment form clients fill in
UpvioThe camera-based body scan, and the patient lookup that supplies the email
n8nThree workflows: form intake, scan intake, and the report email
Python engine (FastAPI)Matching, scoring, AI calls, checks, the job queue and the private report links
MySQL / MariaDBStores assessments, scans, results and the job queue
OpenAIWrites the report narrative from pre-selected findings
WeasyPrintTurns the finished report into a branded PDF
GmailSends the client their report link

The engine runs on the company’s own server next to its database. Clients keep using the same form and scan, and the team keeps working in the tools it already has.

Where This Pattern Applies

This isn’t really about ADHD.

Worked example: a recruitment firm. A candidate fills in a skills questionnaire and records a task on another platform. The two are matched by email, the questionnaire is scored against the firm’s rubric in code, AI writes a candidate summary from the scores only, the summary is checked for banned claims, and a PDF goes to the hiring manager.

The shape: separate inputs are matched to one person, code scores them, AI writes within fixed limits, and checks decide what gets sent. What changes from case to case is the questions, the scoring rules, the second data source and the list of claims that must never appear.

Page · Plain text

Salesforce Marketing Cloud to GoHighLevel Migration & Two-Way CRM Sync

Automated CRM migration, customer journey rebuilding, and real-time two-way data synchronization.

Section: n8n Automations

1. Overview

My Digital Money runs Cash and IRA investment accounts and grows its list with a free crypto guide. Its customer journeys ran in Salesforce Marketing Cloud, driven by Salesforce leads, accounts and opportunities.

The journeys and contact history were moved into GoHighLevel, and n8n now keeps leads, guide requests and opportunities in step between the two CRMs.

Implementation summary:

2. The Problem

The objective was to move customer journeys and contact history out of Marketing Cloud and keep Salesforce and GoHighLevel in sync without manual data entry.

The migration involved six main challenges:

  1. A hard cut-off: The Marketing Cloud subscription was ending, so journeys, email history and unsubscribes had to be exported before access closed.
  2. Journeys wired to Salesforce data: Entry rules and splits read Salesforce fields such as account type (Cash or IRA), funding type and opportunity stage.
  3. Two CRMs, no connection: Salesforce stayed the system of record, but GoHighLevel couldn't tell when a lead arrived, asked for the guide or moved through a deal.
  4. Silent sync failures: Salesforce outbound messages first hit a test webhook URL, then kept retrying because the reply wasn't in the SOAP format Salesforce expects.
  5. Records that had to match: An opportunity could arrive before its contact existed in GoHighLevel, and repeated updates risked creating duplicates.
  6. A brand-new sending domain: mail.mydigitalmoney.com had no sending history, so volume had to start at around 1,000 contacts and grow in stages.

3. How It Works

The implementation consists of six connected components, covering journey migration, data migration, CRM synchronization and email sending.

Step 1 — Journey Audit and Rebuild

Each Marketing Cloud journey was mapped, including its entry source, criteria, splits, emails and waits. It was then rebuilt as a GoHighLevel workflow with the same branches.

Application submitted, IRA

Image: Marketing Cloud: New IRA applications split by initial funding type into two application emails.

Image: GoHighLevel: Contact-created trigger with the same funding-type condition, emails and waits.

Application submitted — Cash

Image: Marketing Cloud: One path: application-received email, a one-day wait, then exit.

Image: GoHighLevel: Contact-created trigger for unapproved individual (Cash) accounts, then the same email and wait.

Step 2 — Data Migration

Contacts, Marketing Cloud email tracking and unsubscribe data were exported, merged and imported into GoHighLevel.

Image: Bulk import: 18.48K records, 100% success, 0 errors.

Image: Contact fields: Salesforce data such as approval, funding type and guide status.

Step 3 — Lead Synchronization

A Salesforce flow sends an outbound message to n8n whenever a lead is saved. n8n reads the full lead and creates or updates the contact in GoHighLevel.

Image: Workflow: webhook or lead trigger → full lead → field mapping → create or update.

Image: Executions: recent runs succeeded in about a second.

Step 4 — Guide Requests

When a lead asks for the crypto guide, n8n waits a minute for Salesforce to finish saving, re-reads the lead, sets Send Guide in GoHighLevel and resets it in Salesforce.

Image: Marketing Cloud: Splits new leads by campaign before sending the guide emails.

Image: GoHighLevel: Send Guide set to true starts the guide email, then the nurture sequence.

Image: GoHighLevel: MDM – Lead Nurture Sequence, shown in three parts

Image: Workflow: wait → re-read lead → set in GoHighLevel → reset in Salesforce.

Image: Executions: each run includes the one-minute wait

Step 5 — Two-Way Opportunity Synchronization

Opportunity changes in Salesforce create or update the matching opportunity in GoHighLevel, and the GoHighLevel ID is written back to Salesforce.

Image: Workflow: update or create in GoHighLevel, then write the ID back to Salesforce.

Image: Executions: successful runs, including batches of updates.

Step 6 — Sending Domain and Warmup

mail.mydigitalmoney.com was authenticated and set as the sending domain for every email type in GoHighLevel.

Image: Synced opportunity: stage mapped from Salesforce, with the Salesforce IDs stored (blurred)

Image: Sending domain: mail.mydigitalmoney.com handles all seven email types.

4. The Result

The migration moved the customer journeys and contact history into GoHighLevel while establishing synchronization between Salesforce and GoHighLevel through n8n.

Seven GoHighLevel workflows now cover the guide, nurture, application, approval, abandonment and first-trade journeys. The main contact import processed 18,480 records with 100% success and no errors.

Leads, guide requests and opportunities now sync through n8n. Recent lead sync runs shown in the dashboards completed in about a second, and opportunity records are linked between both systems using their respective IDs.

The new sending domain was authenticated and used for all seven email types, with a staged warmup process.

5. Results & Metrics

The following figures are taken from the GoHighLevel and n8n dashboards. They represent the documented implementation results, not independent performance audits.

MetricDocumented result
Marketing Cloud journeys documented9
GoHighLevel workflows covering customer journeys7
Contacts in the main import18,480
Main import success rate100%
Import errors0
n8n sync workflows in production3
Emails delivered in the last 30 days120,362
Email bounce rate1.2%
Spam complaints8
Recent lead sync execution timeAbout 1 second

6. Tech Stack

TechnologyPurpose
Salesforce Sales CloudSystem of record for leads, accounts and opportunities.
Salesforce Marketing CloudOriginal customer journeys and email history.
n8nSelf-hosted synchronization between the two CRMs.
GoHighLevelRebuilt workflows, email, SMS and pipeline management.
LeadConnector APIReads and writes GoHighLevel records.

7. Where This Pattern Applies

This pattern is useful for organizations moving from one CRM or marketing automation platform to another while maintaining customer history, automated journeys and reliable synchronization between their systems.

Page · Plain text

End of part 4 of 7. Continue with part 5.