My Device Configuration
CA Firm CRM — Multi-Agent Practice Management System
An end-to-end practice-management platform for CA (accounting) firms, built as a 152-node n8n workflow backing an HTML/JS dashboard front end.
The system is structured around three intake channels — WhatsApp (via Meta's WhatsApp Business webhook, with full verification-token handshake and signed payload parsing), inbound email (Gmail trigger), and a website lead-capture form — each normalized into a shared Google Sheets data layer covering clients, leads, documents, compliance calendar, and invoices.
Two purpose-built LLM agents (backed by Groq) sit on top of this data: a Support Agent that answers client queries on WhatsApp/email using live context assembled from that client's compliance status and document records (with per-channel conversation memory), and a Lead Qualification Agent that scores and classifies inbound leads in real time, auto-escalating hot leads to the partner via Telegram.
Beyond real-time handling, five independent daily schedule triggers run the firm's follow-up operations autonomously: compliance-deadline reminders, pending-document chasers, cold-lead follow-ups, and overdue-invoice reminders — each drafted by an LLM chain grounded in the specific record's data, sent over WhatsApp and email, and auto-escalated to the partner via Telegram once a configurable severity threshold (days overdue / follow-up count) is crossed.
Payment confirmations and document submissions are reconciled automatically against the Sheets records to close the loop without manual data entry. A dedicated error-handling workflow catches failures at any node, logs them to a sheet, and alerts the partner via Telegram — giving the system basic production observability.
How long did it take you to build?
It took me around 5 weeks to build the entire system end-to-end, including the workflow architecture, integrations, AI agents, automation logic, dashboard, and error-handling layer.
How much would you charge for creating something like this?
For a system of this complexity, with multiple communication channels, AI agents, automated follow-ups, lead qualification, dashboard functionality, and production-level error handling, I would charge somewhere in the range of $2,000–$2,500, depending on the firm's specific requirements and level of customization.
04 Aug 2026
Keywords
AI Usage
AI agents handle every client-facing conversation (WhatsApp and email support), lead scoring/classification, and the drafting of all outbound follow-up messages (compliance, documents, leads, invoices) — each grounded in live client records pulled from Google Sheets rather than static templates. I designed the full system architecture manually: the three-channel intake model, the Google Sheets schema tying clients/leads/documents/compliance/invoices together, the five independent daily automation schedules, the escalation thresholds that trigger partner notification, and the workflow-level error-handling and logging layer.
