Case study: AI workflow automation
CRM deal stage automation with compliance checks
Deals moved through CRM stages automatically, with compliance checks, supplier records and QuickBooks hand-offs at each step.
Overview
A sales operations team needed help with this: deals moved through CRM stages automatically, with compliance checks, supplier records and QuickBooks hand-offs at each step.
The problem
Deals had to move through several compliance stages before payment. Someone refreshed the CRM pipeline throughout the day to spot stage changes, checked vendors in QuickBooks by hand, and created follow-up tasks from incoming emails one at a time.
What we built
We built a set of automations around the CRM pipeline that watch for stage changes, run the compliance hand-offs, check vendors in QuickBooks and turn incoming emails into leads and tasks. Each automation is described below.
The automations we built
Each automation below handles one part of the workflow. Together they replace the manual steps described above, and each one keeps a record of what it did.
AML Deal Stage Monitoring
The problem: This automation was built for a sales operations team that needed a reliable way to catch every deal moving into AML status without relying on someone to keep refreshing the pipeline throughout the day. Before this, deals entering AML review could sit unnoticed for a while before anyone realized they needed billing and compliance follow-up, since checking for stage changes wasn't something a person could do continuously. Now the system checks repeatedly throughout the day for deals in AML, sorted by how recently they changed stage, and automatically hands each one off to the next stage of the process for billing and notification.
How it works: The workflow starts by checking for existing bills to avoid duplicate work, then loops through up to 100 checks with a short delay between each pass, continuously scanning for deals that have moved into AML status. It pulls the deals sorted by when they last changed stage, and a router decides whether to keep checking, stop the scenario, or handle an error and wait before retrying. Once a qualifying deal is found, the details are pulled, converted into JSON, and sent through a webhook to trigger the next scenario, which handles the actual billing and notification steps.
- Continuously monitored the pipeline for deals entering AML status
- Eliminated manual pipeline checks throughout the day
- Reduced delays before billing and compliance follow-up began
- Added retry and error handling to keep monitoring running reliably
Vendor Verification Sync
The problem: This automation was built for a sales operations team that needed a reliable way to move deals through AML approval and into payment handling without someone manually tracking stage changes and checking QuickBooks for the right vendor each time. Before this, once a deal was approved through AML, moving it to the payment team and confirming the correct vendor existed in QuickBooks meant several manual handoffs, which slowed things down and made it easy for a deal to stall between stages. Now the system watches for stage changes in Zendesk Sell, routes deals based on ownership once AML is approved, and checks QuickBooks to confirm the right vendor exists before logging the result.
How it works: The workflow starts by watching for deals entering a new stage in Zendesk Sell, then checks whether the stage reflects AML approval before routing the deal toward the payment team. Once a deal is approved, it's routed again based on ownership, and QuickBooks is searched to confirm whether a matching vendor already exists. If a vendor is found, QuickBooks is updated accordingly; if not, the flow logs the result to Google Sheets instead, keeping a clear record of vendor matches for both ownership paths.
- Automated deal handoff from AML approval to the payment team
- Eliminated manual vendor checks in QuickBooks at each handoff
- Reduced delays between AML approval and payment processing
- Kept a clear vendor match record across both ownership paths
Smart Zendesk Lead Creation
The problem: This automation was built for a sales team that needed a consistent way to turn incoming emails into follow-up tasks in their CRM without manually checking each message and creating records by hand. Before this, someone had to read incoming emails, search for the matching contact, and decide whether to create a new lead or update an existing one, which was slow and easy to fall behind on during busy periods. Now the system watches the inbox, checks whether the sender already exists as a contact, and automatically creates the right task or lead in Zendesk Sell based on what it finds.
How it works: The workflow starts by watching Gmail for incoming emails, then searches Zendesk Sell to see if the sender already exists as a contact. If a match is found, the flow checks whether the contact's name needs updating and either updates the person's details or creates a new task, depending on that outcome. If no contact is found at all, the system creates a new lead instead, and a short delay is built in near the start to help pace the process and avoid overlapping runs.
- Automated task and lead creation from incoming email contacts
- Eliminated manual searching for existing contacts before follow-up
- Reduced missed follow-ups on incoming sales emails
- Kept contact records updated automatically as new emails arrived
Related
This project is an example of ai workflow automation. If your team has a similar workflow, the pages below explain how we approach it, and the neighbouring case studies show the same ideas in other businesses.