How a Civil Engineer Built an AI Workforce

We break down how Paul Doherty built three AI agents to replace tasks that would cost $150,000 a year to hire for - and why most non-tech business owners are still missing the opportunity sitting in front of them.

Most non-tech business owners have a quiet, persistent feeling that something is shifting underneath them. They watch every LinkedIn post about AI and quietly wonder whether it applies to them, or whether it's all hype dressed up as opportunity. Most decide it isn't for them. Most decide wrong. The founders who are quietly winning with AI right now aren't running tech startups. They're running construction firms, accountancies, civil engineering practices - businesses where the average team member won't sign a paper time sheet without complaint, let alone adopt an AI workflow.

Paul runs PD Project Delivery, a civil engineering and project management consultancy in Brisbane that operates across Australia. Eight months ago he had never used Claude. Today he runs three AI agents - Hector, Chapo, and Pablo - that handle his daily briefing, HR onboarding, fleet management, and tasks that would otherwise require a full-time hire at $150,000 a year. The first agent took him six to eight hours to build, $1,200 in hardware, and zero coding knowledge. The cost savings compound every week. The team resistance was real, and he learned how to navigate it without losing them.

In this episode, Paul breaks down exactly how non-tech business owners can deploy AI without alienating their team, when to build agents yourself versus pay someone, what tasks actually compound when automated, and what every founder needs to know about the gap opening up right now between business owners who are deploying AI and the ones who are still debating it.

In This Episode, You'll Learn:

How Paul built three AI agents for under $1,500 with zero coding background.
Why AI replaces tasks not jobs - and how to identify the right tasks to automate first.
How to introduce AI to a team that thinks you're trying to replace them.
The cash flow trap most growing founders walk straight into - and how to spot it before it breaks you.
Why selling a business with AI infrastructure is more attractive to buyers, not less.

About Paul Doherty

Paul Doherty is the Founder of PD Project Delivery, a Brisbane-based civil engineering and project management consultancy serving infrastructure projects across Australia.

Originally from Scotland, Paul moved to Australia in his twenties and was sponsored within six weeks of arriving. He worked his way up through senior positions on infrastructure projects across Queensland and New South Wales before founding PD Project Delivery in 2018 with no business plan, no scaling strategy, and a single contract. Seven years later, the firm operates from the Northern Territory to South Australia and delivers some of the largest civil and infrastructure projects in the country.

Today Paul is one of the few non-tech business owners in Australia openly sharing how he's deployed AI inside a traditional industry - building, breaking, and rebuilding the systems he now teaches other founders to use.

Connect With Paul

You Ask, We Answer

Frequently Asked Questions

How Do You Start Using AI In Your Business?

Start with Claude or ChatGPT for low-stakes tasks like drafting emails, summarising documents, and reviewing data. Don't try to automate everything in week one. Most founders see the biggest early wins from offloading the small two-minute tasks they do fifty times a day. Once those are running, build toward agents.

What Are AI Agents And How Do They Work?

An AI agent is a custom-built assistant that runs continuously and handles specific tasks for your business - like an HR coordinator or fleet manager that works 24 hours a day, costs nothing extra after setup, and never asks for time off. They're typically built using tools like Claude, configured to a specific role, and integrated with your business systems through simple automation.

How Much Does It Cost To Build An AI Agent For A Business?

The cost ranges from around $1,200 if you build it yourself (a Mac Mini plus your own time) to roughly $5,000 if you hire someone. Six months ago that price was $7,000 to $9,000, so the market has compressed quickly. The bigger cost is your time investment - expect to lock yourself in for half a day to a full day per agent.

How Do You Get Your Team To Accept AI?

Read the room before talking about it. Most team members hear AI and assume their job is on the line. Show real examples that save your senior people hours of work, not abstract demos. Make it about offloading the low-value tasks they hate, not about replacement. Give them access early - the productivity gains do the persuading.

What Tasks Should You Automate First With AI?

Start with repetitive admin work - data entry, spreadsheet building, report formatting, daily briefings, vehicle and asset tracking, onboarding checklists. The rule is simple: if a computer is punching things out, AI can do it. Save the strategic, relational, and judgement-heavy work for humans. The goal isn't to remove people. It's to free them to do work that actually matters.

Will AI Replace Jobs In My Business?

AI will replace some tasks and probably some basic administrative roles in the next few years. It won't replace roles that require human judgement, relationship management, or accountability. The honest framing: AI will eliminate the parts of jobs that nobody actually wanted to do, and create new roles that don't exist yet. Founders who deploy early write the new job descriptions.

How Do You Use AI In A Non-Tech Industry?

The same way tech founders use it, with different applications. Construction firms use it for handover documentation. Civil engineering practices use it for site reporting. Accountancies use it for client briefings. The industry doesn't matter. What matters is whether you have repeatable processes generating data. If you do, AI compounds on them.

What Is The Difference Between Claude And ChatGPT For Business?

Both are general-purpose AI assistants. Claude (made by Anthropic) is generally stronger at long-form reasoning, nuanced writing, and structured analysis. ChatGPT (made by OpenAI) has broader integrations and more consumer features. Many business owners use both - Claude for deeper analysis, ChatGPT for general queries. The best one is the one you actually open.

Does AI Increase Or Decrease The Value Of Your Business When Selling?

AI infrastructure makes a business more attractive to buyers, not less. Buyers want businesses that can run without the founder. AI agents, documented processes, and automated systems demonstrate the business is structurally sound rather than founder-dependent. Founders deploying AI early are also compressing their exit timelines by years.

How Do You Manage Cash Flow As A Growing Founder?

Treat cash flow as your most important metric, not revenue. Always maintain a buffer covering at least one payroll cycle without incoming receipts. Never assume that doing a good job for someone means they'll pay on time. Chase overdue invoices weekly, not monthly. And separate profit and loss from actual cash position. They aren't the same thing, and confusing them is how growing businesses go broke.