Why treating AI like your friend could cost your business

Boxers & Briefs Podcast #36: The future of AI and cybersecurity and what it means for businesses with Claude Chiorean

Everyone’s talking about AI, but most businesses are getting the relationship wrong from the start. They’re approaching artificial intelligence like it’s a helpful colleague rather than what it actually is—a powerful tool that requires careful management and control.

Claude Chiorean, managing director of ‘WE AS WEB New Zealand’, has spent over a decade in cybersecurity and now helps businesses implement AI solutions safely. His perspective cuts through the hype with a simple analogy: “I don’t trust my car. It’s not my friend. I’m using my car for an outcome, and I control my car.”

The same principle applies to AI. It can transform your business operations, but only when you maintain proper oversight and understand its limitations.

Building bridges across skill gaps

Claude’s journey to AI consultancy began with a problem many New Zealand businesses face—the resource gap. After years working as a cybersecurity consultant for multinational companies, he noticed two consistent complaints from clients: budget constraints and lack of skilled personnel.

“Budget is budget—every company has their own financial plans,” Claude explains. “But in terms of resources, there’s always been a problem. We don’t have enough people, there’s a skill gap.”

The solution came through partnership with a European company that employs around 600 developers across eight offices. WE AS WEB New Zealand launched a year ago as a local entity backed by this extensive overseas resource pool.

“We can resource them tomorrow,” Claude says. “If a customer needs one developer for a small project, we can help. If they need seven developers for a bigger project, we can provide them basically overnight.”

This model addresses the reality that many New Zealand businesses want to implement new technology but lack the in-house expertise to do it safely and effectively.

Security first approach to outsourcing

When businesses outsource development work, particularly to overseas teams, security becomes paramount. Claude’s cybersecurity background shapes every aspect of how WE AS WEB operates.

“We vet all professionals that work with us—they sign NDAs and legal contracts,” he explains. The process includes careful onboarding where developers receive exactly the access they need to complete their work, nothing more.

Perhaps more importantly, the offboarding process ensures no residual access remains when projects end. “When the professional stops working on that specific project, intellectual property belongs to the customer. There’s no residual code in our professionals’ environment.”

All file transfers are encrypted, and the company maintains what Claude calls a ‘cybersecurity checklist’—procedures developed through hard-won experience in the field.

When ransomware locks down government

Claude’s most sobering cybersecurity story involves a government agency that suffered a devastating ransomware attack. Someone clicked on a malicious email link, giving hackers access to spread throughout the network and lock down critical systems.

“One of the services locked down was their payment system,” Claude recalls. “Imagine a government agency frozen—they cannot pay, cannot receive money, cannot transfer money. Everything stops.”

The hackers hadn’t stolen money—they’d simply encrypted the systems, making them inaccessible. The ransom demand was in the millions. But perhaps worse than the initial attack was discovering the hackers had also found and encrypted the backup systems.

This story illustrates a fundamental cybersecurity truth: “They have to get it right once to get into the system. We have to get it right every time because we have to stop them.”

The asymmetry is stark. Cybercriminals can afford to make mistakes because they simply move to the next victim. Businesses must maintain perfect defences because one breach can be catastrophic.

AI as tool, not friend

As businesses increasingly look to AI for solutions, Claude advocates for a measured approach grounded in reality rather than hype.

“We have to be careful and not think that AI is our friend,” he emphasises. “AI is not our friend. AI is a tool that we need to use for our benefit.”

He compares AI to a car—useful for getting from point A to point B, but requiring control and maintenance. “I don’t trust my car. It’s not my friend. I’m using my car for an outcome, and I control my car.”

This perspective becomes crucial when businesses consider implementing AI solutions. The technology can provide significant benefits, but only when properly managed and understood.

Practical AI implementation for small business

Despite the risks, Claude sees tremendous opportunity for small and medium businesses to leverage AI effectively. The key is starting with specific pain points rather than trying to implement AI for its own sake.

“We sit down with customers and say, ‘Tell me the top five or ten pain points you have in your business,’” Claude explains. These might include sales processes, data analytics, marketing, or customer support challenges.

“About 80% of those can be automated,” he claims. The most common example is customer service, where businesses find themselves answering the same basic questions repeatedly.

Traditional chatbots frustrate users because they follow scripts, creating dead ends when customers ask questions outside predetermined parameters. Modern AI-powered chatbots can have actual conversations based on company data and documentation.

“Instead of just following a script, the chatbot can have access to lots of data and respond with appropriate answers based on that data,” Claude explains. “It’s conversational rather than scripted.”

For a small five-person business, implementing such a system might cost a few thousand dollars rather than the tens of thousands many assume AI requires.

Data analytics and financial forecasting

Beyond customer service, AI excels at pattern recognition in financial data. Small businesses can feed years of financial information into AI systems to identify trends and opportunities.

“It can tell you what your best and worst months are going to be based on historical trends,” Claude notes. More importantly, it can analyse which products or services generate the highest profit margins and suggest focus areas for maximum return.

WE AS WEB is developing a subscription-based ‘AI as a service’ platform specifically for financial analysis. The system will connect to existing financial software like Xero or MYOB and provide instant insights through a conversational interface.

“A CFO could say, ‘Hey Alex, I want to know my profit and loss for the last month. Put it in PowerPoint format and send it to my email. I have a meeting in ten minutes,’” Claude explains. “Boom—you’ll get it in your inbox.”

Security concerns with cloud-based AI

Many businesses hesitate to put sensitive financial data into AI systems, particularly cloud-based platforms where data processing happens in ‘black boxes’ with little transparency.

Claude acknowledges these concerns while pointing toward emerging solutions. ‘Edge AI’ involves downloading pre-trained AI models to run locally on company devices rather than sending data to external cloud services.

“You can actually download and install pre-trained AI models on your laptop,” he explains. “You get the same results, but everything stays local.” These models require periodic updates but don’t send sensitive company data to external servers.

This approach addresses privacy concerns while still providing AI capabilities. The technology exists now through platforms like Hugging Face, which offers over 550,000 downloadable AI models.

The explainable AI revolution

The strawberry counting error represents a broader challenge with current AI systems – they provide answers without showing their reasoning. ‘Explainable AI’ represents the next evolution, where systems not only provide answers but explain their logic.

“Instead of sending something into a black box and not understanding how that AI gave you the answer, explainable AI will give you the answer and explain how it reached that conclusion,” Claude says.

This transparency becomes crucial for business applications where understanding the reasoning behind recommendations matters as much as the recommendations themselves.

Healthcare transformation on the horizon

Looking ahead, Claude sees healthcare as the sector most likely to experience dramatic AI-driven transformation. He points to medical pods already operating in San Diego, where people pay around $99 per month for comprehensive body scanning and analysis.

“You sit in a chair, scan your body, and they check blood pressure, diabetes risk, and lots of other things,” he explains. These systems will evolve toward preventative medicine, identifying health risks before symptoms appear and providing personalised treatment plans.

“Instead of everyone with a headache getting the same treatment, the system might say you have a headache because you’ve had too much sugar or not enough water,” Claude suggests. This personalised approach could revolutionise healthcare delivery and outcomes.

Transportation and autonomous systems

Self-driving vehicles represent another major AI application gaining real-world traction. Claude describes electric trucks that have been running autonomous routes between Los Angeles and New York for five years, learning from accidents and problems to continuously improve.

“They learn and learn and learn,” he says. “Now they’re evolved enough that they’re putting trucks on roads to deliver goods in Texas—actually working and delivering groceries from warehouses to supermarkets.”

This progression from controlled testing environments to real-world applications demonstrates how AI systems mature through experience and iteration.

Starting small with measurable outcomes

For businesses ready to implement AI, Claude’s advice centres on starting small and measuring results carefully. Rather than attempting comprehensive AI transformation, companies should identify one significant pain point and solve it thoroughly.

“Once that is implemented and working and gives you measurable outcomes, then you can take a percentage of that saving and put it into the next problem,” he suggests.

The emphasis on measurement cannot be overstated. “Sometimes people say, ‘I’ve implemented AI and it’s working fine.’ Okay, but did you save money? Did you save time? Measure it.”

Without clear metrics, businesses risk implementing AI for novelty rather than value. The goal should always be quantifiable improvements in efficiency, cost reduction, or revenue generation.

Finding the right partners

Small and medium businesses face a fundamental challenge with AI implementation – they typically lack the internal expertise to develop and deploy these systems safely and effectively.

“We’re talking about AI developers, prompt engineers, people who actually know what they’re doing,” Claude notes. “They are very expensive. You cannot hire someone like that without spending a lot of money.”

This reality makes partnering with experienced AI consultants more practical than building internal capabilities from scratch. The key is finding partners who understand both the technology and the specific challenges facing smaller businesses.

Avoiding the hype trap

Claude’s measured approach to AI contrasts sharply with much of the current market hype. While acknowledging AI’s transformative potential, he emphasises the importance of realistic expectations and careful implementation.

“Don’t think that AI is going to solve all your problems overnight,” he warns. “It’s a tool that requires proper management and control.”

This perspective becomes particularly important as businesses face pressure to implement AI quickly to avoid being left behind. The reality is that successful AI implementation requires careful planning, proper security measures, and ongoing management.

Preparing for an AI-driven future

The conversation around AI often focuses on dramatic future scenarios, but Claude’s approach emphasises practical steps businesses can take today. By starting with specific problems, implementing solutions carefully, and measuring results thoroughly, companies can build AI capabilities gradually while managing risks appropriately.

The technology will continue evolving rapidly, but the fundamental principles remain constant. AI works best when treated as a powerful tool requiring human oversight, not as an autonomous solution that operates independently.

For New Zealand businesses, the opportunity lies not in competing with tech giants or implementing cutting-edge AI research, but in applying existing AI capabilities thoughtfully to solve real business problems. The companies that succeed will be those that balance ambition with pragmatism, embracing AI’s potential while respecting its limitations.

As Claude puts it, AI isn’t about replacing human judgment—it’s about augmenting human capabilities with tools that can process information faster and identify patterns we might miss. The key is maintaining control of the process while leveraging the technology’s strengths.

This article and podcast is proudly brought to you by Gilligan Sheppard, the problem solvers in business who believe in thinking differently.