Yunus Mansuri
AI Workflow Automation | n8n
New to Contra
Yunus is building their profile!
šØ What if your sales team only got alerted when a deal actually became risky?
I built a RevOps Deal Risk Assessment & Alert System in n8n to solve a problem that many sales teams face:
Too many deals.
Too many status checks.
Too many repetitive notifications.
And not enough visibility into which changes actually matter.
š The idea
The workflow connects HubSpot ā n8n ā Google Sheets/Data Tables ā Slack and evaluates deals on a scheduled basis.
Instead of simply checking whether a deal is "high risk" and sending another alert every day, the system tracks what happened previously and decides whether the current change actually requires action.
Each deal is routed into one of four outcomes:
šØ Alert ā meaningful risk detected
š Silent Refresh ā state changed but no notification is needed
ā
Resolution ā previously risky deal has recovered
ā No Action ā nothing meaningful changed
š§ What I focused on
The interesting part wasn't connecting the apps.
It was making the automation reliable.
I designed:
ā¢ Persistent state management
ā¢ Alert deduplication
ā¢ Delivery-confirmed state commits
ā¢ Retry-safe notification handling
ā¢ Deterministic risk scoring
ā¢ AI-assisted alert wording with a fallback
ā¢ Centralized error aggregation
ā¢ Reusable error handling
ā¢ Detailed workflow documentation
For example, the system only records an alert as delivered after the messaging platform confirms delivery.
So if the notification fails, the workflow doesn't incorrectly assume that the alert was already sent ā it can try again on the next run.
š¤ Where AI fits
I deliberately didn't let AI make the core risk decision.
The actual risk assessment uses deterministic business rules so the result remains predictable and explainable.
AI is used only where it adds value: generating a concise, human-readable explanation for the alert.
That means an AI failure doesn't remove the underlying risk signal.
š Built with
n8n ā¢ HubSpot API ā¢ Google Sheets/Data Tables ā¢ Slack ā¢ Gemini
This project started as a portfolio exercise, but the bigger goal was to practice designing automations that are not just functional ā but stateful, recoverable, explainable and maintainable.
#n8n #WorkflowAutomation #RevOps #SalesOps #Automation #AI #HubSpot #NoCode #LowCode #BusinessAutomation
0
28
Built an end-to-end RevOps automation in n8n that evaluates open HubSpot deals every morning, scores deal risk using deterministic business rules, explains the drivers behind each risk, maintains an operational action queue, and sends Slack alerts only when a material change occurs.
0
51
Built an automated lead classification and inbox triage pipeline. The system uses Gemini 2.5 Flash to analyze incoming emails for intent, urgency, and sentiment. It automatically routes high-priority leads to instant alerts, generates pre-written support drafts for customer service, and logs all data in Google Sheets . Built with multi-step routing and fallback error handling to eliminate manual inbox management and accelerate response times.
0
37