Construction AI Support Agent with Human Handoff

A regional construction company was losing leads because site inquiries, material questions, and scheduling requests were piling up in shared inboxes. I built an AI agent that handles the routine questions automatically and pulls a human in only when a project truly needs it.

Inbound messages were spread across email, WhatsApp, and the website contact form. The team had to:

• Manually read and triage every inquiry
• Repeat the same answers about materials and timelines
• Forward project-specific questions to the right manager
• Lose leads when responses came in too late

This worked, but it slowed down response time and added unnecessary manual work.

I built an AI support agent connected to the company's channels, documents, and CRM.

1. 1Answers common questions about services, materials, and timelines
2. 2Collects project type, location, budget, and contact details
3. 3Saves the conversation and creates a CRM record
4. 4Hands off to a human project manager when the inquiry is complex or high-value
5. 5Notifies the team in real time so no lead is missed

• n8n
• OpenAI
• WhatsApp Business API
• Google Sheets
• Slack

The agent receives the message, classifies the intent, and either answers from the company's knowledge base or asks structured questions to qualify the lead. Once enough detail is captured, it saves the record and notifies the right project manager with full context so they can take over the conversation seamlessly.

After launching the agent the team saw immediate impact:

The team now spends their time on actual project work instead of triaging inbound messages.

The biggest win wasn't the AI itself — it was designing the handoff. Knowing exactly when to stop answering and pull in a human is what made the team trust the system.