AI system for reporting and monitoring health deals for Zoho CRM - daily review for CEOs and weekly spot diagnosis of risk deals.



Project details

Context:

Problem:


Solution

<aside> 🧠

Two complementary AI processes: a daily portfolio-wide report for leadership, and a weekly deep-dive specifically on delivery-stage deals — each pulling full call/task/note history before scoring, and alerting only when something actually needs attention.

Tools used: n8n, Zoho CRM, Claude (Opus 4.5 + Sonnet 4.6), Google Sheets, Telegram.

</aside>

Solution logic:

  1. Daily report (9:10 AM): pulls all deals from Zoho CRM, aggregates stats (delivery/feedback counts, amount at risk, stage distribution) with code, then an AI Agent (Claude Opus 4.5) decides which specific deals warrant a closer look and calls a CRM tool per deal ID to diagnose issues — producing a prioritized Telegram report for the CEO.
  2. Weekly Deal Health Alerts (Mon 10 AM): loops through every delivery-stage deal one at a time, pulls its full history (calls, tasks, notes) via a reusable sub-workflow, and has Claude Sonnet 4.6 score the deal's health 1-10 with named risk factors and a recommended action.
  3. Results are logged to a dedicated "Pipeline" tracking sheet; Telegram alerts fire only for critical/medium-risk deals — healthy deals are silently skipped so the channel doesn't turn into noise.