Latest project · AI automation

Client: Canada's Log and Wood Home Store

AI-powered lead follow-up, with people in control.

AKNEX Digital turned live chat and contact-form enquiries into one reliable workflow for context-aware AI drafts, human approval, customer outreach, and lead tracking.

AI-assisted responses Human approval Centralised lead tracking
Overview of the AI lead follow-up and response management system
The challenge

Customer conversations needed a faster, more dependable path to follow-up.

Canada's Log and Wood Home Store needed to respond quickly without giving AI unrestricted control over customer communication. Each draft had to understand the full conversation, reflect the business's knowledge, and remain subject to staff review.

  • Bring live-chat conversations and website enquiries into one workflow.
  • Alert staff without requiring constant dashboard monitoring.
  • Preserve multi-turn conversation context and prevent duplicate processing.
  • Log customer and lead information reliably for future management.
  • Learn from approved responses, edits, and client feedback.
The solution

A custom response-management system built around a human approval loop.

AKNEX Digital connected Zoho SalesIQ, a custom Node.js backend, xAI Grok, Google Sheets, email infrastructure, and an administration dashboard into a single workflow.

High-level architecture of the AI lead follow-up solution
One connected workflow

From first message to approved follow-up.

The system keeps the full conversation intact, prepares a business-aware draft, alerts the team, and waits for a person to approve or improve the response before anything reaches the customer.

  1. 01
    Customer enquiry arrives

    Zoho SalesIQ receives the live-chat conversation and webhook events.

  2. 02
    Conversation is processed

    The custom Node backend assembles messages, visitor details, and the full conversation state.

  3. 03
    AI prepares a suggested response

    xAI Grok analyses the transcript against business knowledge and sales instructions supplied by Canada's Log and Wood Home Store.

  4. 04
    The team is notified

    An email contains the customer's details, transcript, enquiry, suggested response, and relevant lead information.

  5. 05
    A person reviews and sends

    Staff can approve the draft or edit it in the rich-text editor before customer outreach.

  6. 06
    Feedback improves future drafts

    Approved responses and corrections become useful examples for later suggestions.

AI-assisted, human-approved

AI handles the preparation. People keep the final say.

The suggested response considers the current message, previous conversation history, business knowledge from Canada's Log and Wood Home Store, custom sales instructions, and examples approved by the team.

Good responses can be approved immediately. Drafts that need improvement can be edited first, giving the client complete control over what is communicated.

Human-in-the-loop AI supervision for customer response management
Learning from feedback

Every approval and correction makes the next draft more relevant.

The response layer does not rely only on a generic model. It incorporates business-specific knowledge, positively rated drafts, and real corrections from the client's workflow.

Feedback loop from AI draft through human feedback and correction
Administration dashboard

One place to review conversations, manage leads, and send outreach.

The dedicated interface centralises the work that was previously spread across chat, email, spreadsheets, and separate systems.

Conversation management

  • Full transcript inspection
  • Conversation history
  • Actionable and incomplete lead identification
  • Schema repair for inconsistent data

AI response management

  • Context-aware follow-up drafts
  • On-demand regeneration
  • Human editing and approval
  • Exemplar-based learning

Customer outreach

  • One-click email sending
  • Rich-text HTML editor
  • Formatted email generation
  • Bulk lead management

Lead management

  • Google Sheets synchronisation
  • Lead-status management
  • Search, filtering, and pagination
  • Lead deletion and bulk deletion
Reliable conversation tracking

State management prevents duplicate processing and duplicate replies.

Live-chat platforms can emit multiple webhook events for one conversation. AKNEX developed a dedicated state-management layer that understands which messages belong together and when a conversation is ready for processing.

  • In-memory session tracking and two-minute inactivity windows
  • Map/Set-based locks and duplicate-trigger protection
  • Chat reconnection handling and payload normalisation
  • Asynchronous fault isolation across external integrations
Technical flow for session state management and duplicate prevention
Technology and integrations

A custom backend connected the tools already central to the business.

Zoho SalesIQ, Node.js, Express, xAI Grok, Google Sheets, and Nodemailer integration flow
Node.js + ExpressWorkflow engine and API logic Zoho SalesIQLive-chat webhooks xAI GrokContext-aware draft generation Google Sheets API v4Lead and conversation logging Nodemailer + SMTPNotifications and outreach Zoho CRMContact-form lead processing cPanel + Phusion PassengerProduction hosting GitHub ActionsAutomated deployment
Extended beyond live chat

The contact form now enters the same lead-management pipeline.

The site's Contact Us form was connected to a reusable inbound process: payload normalisation, Zoho CRM, Google Sheets logging, an AI follow-up draft, dashboard record, email alert, and human outreach.

This gives both live-chat and website enquiries a consistent path from first contact to follow-up.

Nine-step website contact-form lead pipeline
The result

A centralised, AI-assisted lead-management system.

The new workflow captures incoming enquiries, maintains context, prepares responses, alerts staff, centralises records, and keeps human approval at the centre of communication.

-75%Average response time
+30%Lead conversion rate
100%Lead data captured in the dashboard
Business results of the optimised lead-management workflow
AKNEX Digital's role

End-to-end engineering, from requirements to continuous improvement.

AKNEX Digital handled requirements analysis, architecture, backend development, API integrations, AI prompt engineering, state management, dashboard development, email automation, CRM integration, testing, production deployment, and ongoing enhancements.

AKNEX Digital end-to-end software engineering lifecycle for the project

AI handles the heavy lifting. Humans remain in control.

Discuss your automation project