Technology is moving fast. The builders who lean in now will be the ones still standing when it catches up.

There is a lot of noise around AI right now.

Every second LinkedIn post. Every conference keynote. Every software vendor promising to revolutionise your workflow with a chatbot and a subscription fee.

Most of it is selling you something.

What Drew and Sean from Increase Construction are doing is different. On our latest episode of The Good Builder Podcast we cut through the noise in a way that genuinely matters for anyone running a trade or construction business in Australia.

The core message is simple. AI does two things. It saves you time and it makes you money. If a tool does neither, it is not worth your attention.

The Problem They Were Trying to Solve

Like a lot of growing construction businesses, Increase Construction had reached a point where the workload demanded more hands. The administration burden was real. The client-facing side of the business was being squeezed by the operational side.

For close to a year, Drew and Sean planned to hire. Every time they were ready to pull the trigger, something new emerged in the AI space that gave them pause.

That instinct turned out to be right.

Rather than employing someone to manage reporting, marketing coordination, and operations administration, they built a team of AI agents to do it instead. Not a chatbot sitting in a browser tab. An actual multi-agent system running in the cloud, with named roles, defined responsibilities, and a general manager overseeing the lot.

They called the General Manager Mando.

Mando reports to Drew and Sean. Underneath him sit a Marketing Manager, an Operations Manager, a Sales Manager and an HR Manager. Each has been trained on years of the business’s content, context, and working style. Each operates with enough autonomy to complete tasks without being prompted from scratch every time.

This is not a product pitch. It is the result of months of deliberate, often painful, technical work. Sean broke the system somewhere between 30 and 40 times before it started performing.

What an AI Agent Actually Is

For builders who have not gone deep on this yet, the term “agent” can sound opaque.

The straightforward version is this. An agent is a set of instructions and a defined personality, given context and parameters about how it should operate. Where a standard chatbot waits to be asked something and responds once, an agent can be given ongoing tasks, run scheduled reports, monitor other agents and take action without constant supervision.

The difference between using ChatGPT to draft an email and using an agent system is roughly the difference between a single phone call and a full-time employee.

Sean’s analogy is worth sitting with. The output you get from any AI tool is only as good as the context you feed it. Agents that have been properly trained on your business, your tone, your processes and your goals operate at a fundamentally different level than a generic prompt typed into a browser.

The Platform: OpenClaw and OpenRouter

For the technically curious, the system the team built runs on OpenClaw, installed in a private cloud server rather than on a local machine. The cloud setup costs around 30 to 40 dollars a month and keeps sensitive business data separate from personal devices.

From there, they access AI models through OpenRouter, a platform that provides access to hundreds of different models including Claude, ChatGPT and Gemini. The key insight Sean offers is that different models have different strengths. Claude tends to perform well on document creation and pushes back more critically. ChatGPT handles certain process-driven tasks well. The agent system selects the right model for each task automatically.

Total daily running costs once optimised: around five to ten dollars.

That figure matters for builders trying to understand the actual economics of this. The infrastructure cost is not the barrier.

The Use Cases That Should Interest Builders

The conversation covered several applications that have direct relevance to construction businesses.

Content and marketing is the obvious one. The Increase Construction team fed their marketing agent years of Drew’s writing, LinkedIn posts, and podcast content. When a test post was generated and published to LinkedIn, Drew did not feel the need to edit it. The voice was right. The scale potential from that is significant for builders who know they should be more visible but simply do not have the bandwidth to show up consistently across multiple channels.

Lead generation is another. The team tested an agent that scrapes online directories and portals to compile detailed profiles of trades and sub-trades in specific regions. The test cost them a $2,000 bill when a loop error kept firing 90,000 requests to Google through the night. Sean is candid about that. It is also evidence that the capability is real, and the system works when it is configured correctly.

The use cases that genuinely surprised the room during the podcast were the ones that came from thinking about construction specifically.

An agent trained on the NCC that answers compliance questions for supervisors and site teams. A defects management bot that builders can include in their handover documentation, allowing clients to query whether a specific issue is covered. Condensation management guidance built around the new code requirements, accessible from a phone on site.

These are not hypothetical applications waiting on some future version of the technology. They are buildable now.

Where the Risk Sits

Sean is measured on this, which is part of why the conversation carries weight.

Security is a legitimate concern. The team deliberately keeps financial data separate from the agent system. They are not rushing to give the agents access to client financials or sensitive CRM data. The approach is to give the system the information it needs to do the defined job and nothing more.

The $2,000 Google API bill is a real reminder that autonomous systems, when misconfigured, can move fast in the wrong direction. Sean describes it as a lesson for anyone without a technical background running these tools without guardrails.

There is also an honest warning about the market. The AI space is currently flooded with people selling products that add an AI label to something that was always just a CRM, a reporting dashboard, or a marketing service. The Increased Construction position is direct: before any technology conversation, the question should be what problem you are actually trying to solve and where your bottlenecks sit. Tool-first thinking leads to subscriptions you do not use and implementations that fail quietly.

What This Means for Administration in a Building Business

The shift that Drew describes is not about replacing builders or trades. It is about removing the administrative layer that currently sits between information and decisions.

In a traditional business, a managing director wanting to understand current performance has to wait for an administrator to gather the data, pass it to middle management, compile a report, and schedule time to present it. That process can take days.

With a properly configured agent system, the same information is available on request, structured and current.

The implication for builders is not that they need fewer people. It is that the people they have can focus on what actually moves the needle. Client-facing time. Site relationships. Quality decisions. Trade management. The work that requires a human to be present.

Administration is not eliminated. It is handled.

The Honest Advice on Getting Started

Neither Drew nor Sean oversells the entry point.

Getting this right takes time, technical patience and a willingness to lean in properly. Half-measures produce weak results and a quick return to the old way of doing things. The technology rewards commitment.

Their practical advice is to start with the outcome in mind. Where do you want the business to be in 12 months? What is currently wasting your time? What would you hire someone to do if you could afford it?

From there, either find someone who has done this work before and can build it for you, or hire a technically capable person internally who has the focus and time to learn. The tools are accessible. The expertise required to configure them well is not yet common.

That gap is exactly where Increased Construction sits. And it is a gap that will close quickly as more operators figure this out.

The Bottom Line

The builders and operators who lean into this now will have a meaningful head start. Not because the technology is magic. Because the willingness to figure it out before everyone else does is always how competitive advantage works in this industry.

Save time. Make money.

Everything else is noise.

To hear the full conversation with Drew and Sean from Increased Construction, listen to The Good Builder Podcast. 

Author: TGB Editorial