In conversation: Talking Tech and Artificial Intelligence with John Fitzpatrick, CTO of Nitro

Artificial intelligence (AI) has transitioned from a futuristic idea to a significant influence that has been propelling transformation and innovation in various sectors. The tech marvel has arrived in full force and could usher in the biggest technological shift in generations.

As per reports, the AI market is projected to reach a staggering $1,339 billion by 2030, experiencing substantial growth from its estimated $214 billion revenue in 2024.

We spoke with John Fitzpatrick, CTO of Nitro, to explore this pressing topic further. With expertise in AI Innovation, Tech Innovation, and Computer Science, John brings a comprehensive understanding of AI’s rapid evolution and its potential impact in the future.

Q1. AI-washing is a growing concern. As an experienced CTO, how do you think leaders can avoid this?

Fitzpatrick: AI-washing has become pervasive. Companies exaggerate or misrepresent AI capabilities, often rebranding existing business logic or adding superficial integrations that add minimal real value. We also see this on the people side, with many consultants and IT services rebranding themselves as AI experts despite lacking genuine AI expertise or a track record of shipping AI products.

Today, nearly every company mentions AI in earnings calls, and startups need an AI angle to secure funding. Still, those substantially leveraging AI internally or in their products represent a much smaller group.

At Nitro, we start with the customer’s problem or pain point rather than injecting AI for its own sake. This means we often end up with features that don’t even look like AI—they just seem magical. As the industry matures beyond the hype curve and AI becomes ever easier to integrate, it is becoming an implementation detail for building powerful product features, albeit a mighty and transformational one.

Evaluating a vendor’s AI features or solutions is not very different from assessing any other software or service. A healthy skepticism is paramount—focus on the value it creates and make sure it solves a real business problem.

When considering a vendor that leverages AI, make sure you understand how they have integrated AI and how they handle data, ensuring that your data remains secure.

Ask questions such as, “How exactly does your solution leverage AI?” “Are you using third-party AI services?” and “How do you ensure data security and compliance?” This will often expose limitations or weak privacy practices.

At Nitro, we run our model inference services in dedicated private instances that we control. Furthermore, we do not store customer documents or log private data from customer interactions with our AI services – this allows us to provide our customers with the privacy and security they have come to expect from us. 

Lastly, remain wary of solutions promising complete automation without human oversight. The most effective implementations complement and augment human capabilities rather than replace them.

Q2. Do you think ChatGPT is a success? Has the business landscape happily accepted this tech innovation?

Fitzpatrick: Yes, no question. ChatGPT fundamentally reset expectations about what AI is capable of, demonstrating capabilities that most thought were a decade or more away.

Beyond the technical achievement, it also marked an inflection point for AI in terms of awareness – it thrust AI into everyday conversations fueled by media attention. This, in turn, put pressure on businesses to develop an AI strategy, resulting in the AI washing phenomenon mentioned earlier. 

However, there’s a significant gap between consumer AI experiences and enterprise requirements, particularly for accuracy, security, and specificity. While ChatGPT and other large language models successfully demonstrate AI’s potential to a mass audience, business reception has been nuanced, and it has taken longer for businesses to adopt AI at scale. 

We are now moving beyond the initial hype phase. Companies are learning what AI can and cannot do, which is leading to more pragmatic approaches, practical applications, and better product features. Investors and businesses are getting better at distinguishing between superficial AI integrations and those that add real utility. Over time, companies that fail to successfully leverage and adopt AI will move away slower, have higher costs, and risk being leapfrogged. Ultimately, they will be left behind by their more forward-looking competitors. 

Our customers find value in specific applications like form automation, data extraction, and document summarization — but always with human oversight. These AI-powered features automate administrative and laborious tasks, saving significant time for our users and allowing them to focus on what matters.

ChatGPT’s real success has been the mindset shift it sparked, catalyzing meaningful conversations about governance and value. Sophisticated businesses aren’t asking, “Should we use AI?” but, “Where specifically can AI enhance workflows while maintaining necessary controls?”

Q3. Today, more and more threat actors are relying on AI and GAN to cause cyberattacks. What is your thought process on how these attacks can be controlled or eliminated?

Fitzpatrick: This arms race requires both technical vigilance and process innovation. The next major corporate breach may come from AI crawlers harvesting data via unsecured documents, despite cybersecurity investments. 

One critical defense is implementing privacy-preserving AI approaches. At Nitro, we have a maniacal focus on user privacy and security. Practically speaking, that means using dedicated instances where no customer documents are stored after processing, strict data retention policies, and even avoiding logging how customers interact with our document assistant technology. Minimizing the data footprint reduces the risk of attackers exposing or misusing data.

Another risk is that people within organizations send private company documents to unapproved third-party AI tools. In most cases, these individuals do not intend to harm the organization; they are simply trying to leverage AI to improve their workflow. It is, therefore, important for organizations to understand this need and, rather than fighting it, provide approved solutions from trusted vendors.

It is also important for organizations to implement an internal AI policy specifying approved tools and data usage guidelines to maintain security standards.

For enterprises working with AI vendors, especially in document workflows, it’s essential to ask specific questions about how data is protected, whether they use customer data for training, and if they’re using third-party services that might expose sensitive information. Verify compliance with security standards like GDPR, SOC2, and ISO 27001 certification.

Q4. As a leader, how will you encourage employees to use AI as a tool for guidance & support and not a shortcut that takes away the learning process?

Fitzpatrick: I think of AI as an amplifier of human expertise rather than a replacement. We provide space for our teams to experiment and learn about technologies through internal hackathons and dedicated capacity for learning. 

Within engineering, we’ve tested a range of AI-powered coding tools such as ChatGPT, Claude, GitHub copilot, Cursor, and others–listening closely to our teams to identify and approve the ones that genuinely streamline workflows.

While AI cannot replace the critical thinking and creativity of talented software engineers, it does accelerate their output, particularly for handling boilerplate code, suggesting improvements, and easing access to documentation, freeing them to focus on more complex problem-solving.

I encourage all engineers to understand the user at the end of every line of code. This customer-centered approach naturally guides thoughtful AI implementation because teams recognize where automation adds value versus where human judgment remains crucial.

We’ve found the best engineers can bridge customer pain points to designing well-architected software and AI systems. When building AI teams, it’s important to find people who understand both the capabilities and limitations of the technology and genuinely care about understanding the users.

Q5. How worried should we be that AI will replace jobs? Is AI a boon or a curse?

Fitzpatrick: I see transformation rather than wholesale replacement – AI will reshape jobs, not replace them. 

At Nitro, we’ve found that while AI excels at specific tasks like data extraction and summarization, the full context-aware and nuanced understanding of documents remains distinctly human. 

Jobs centered on repetitive, deterministic tasks are vulnerable. However, for knowledge workers, AI can become a collaborative partner handling mundane aspects, while humans focus on higher-value tasks. Think of it as a junior team member that can process documents to extract the key points and help automate workflows while the human user reviews and makes the high-level decisions. 

As with all technological change, the real concern isn’t about the technology itself but managing transition periods. By embracing the technology, training staff in new AI tools, and emphasizing human-AI collaboration skills, we turn potential job threats into opportunities for growth.

AI is a powerful tool whose impact depends entirely on how thoughtfully we deploy it. With proper governance and human-centered design, it can free us from drudgery while creating new forms of valuable work.

Q6. What advice would you like to give to future tech leaders?

Fitzpatrick: The only certainty is constant change, so embrace continuous learning and cultivate genuine curiosity alongside a healthy skepticism. This balanced mindset will help you make smarter decisions than either cynics or overly enthusiastic early adopters. 

Strive to be T-shaped. The most effective tech leaders combine deep technical expertise with a strong product mindset and excellent people and communication skills. This cross-disciplinary breadth gives you the context to make better decisions, build trust across functions, and lead from the front. This, in turn, enables you to align teams, inspire confidence, and bring entire organizations on the journey with you.

When hiring, look beyond credentials to enthusiasm, intelligence, and willingness to learn. The best engineers are technical experts with a customer focus and extreme ownership. When hiring in the AI space, consider both AI scientists from academic backgrounds and software engineers who’ve upskilled in AI—the right balance depends on how you’re applying AI in your business.

Always put the user at the heart of development. AI is only as good as the experience of the person using it. This human-centered approach naturally guides responsible innovation and helps engineers understand the real-world problems they’re solving.

Finally, embrace the responsibility of developing AI systems. Tech leaders now must consider not just what the technology can do, but what it should do and how it should do it. It is important to set high ethical standards for yourself and your teams, putting the user and their privacy at the forefront of all your decisions. Be transparent with customers and stakeholders about how your AI or technology works and the measures in place to use it responsibly.

Build trust through transparency, not speed. Rushing to market with solutions that compromise on quality, privacy, or fairness might win short-term attention, but it won’t earn long-term success.

Bio:

John Fitzpatrick is an accomplished technology executive with a track record of shipping innovative products across AI, SaaS, cybersecurity, and telecommunications. As Chief Technology Officer (CTO) at Nitro, he oversees the company’s technology strategy, product innovation, and engineering teams.

Before joining Nitro, John was at Apple, where he led the development of Apple’s NL Tooling and Operations platform from its inception. This platform enabled the training and deployment of privacy-preserving on-device AI models across more than 1 billion Apple devices in over 40 languages. Additionally he spearheaded the initial model training and internal hosting platform for Apple Intelligence models.

Prior to Apple, John was VP of Product & Engineering at Voysis, a voice AI company acquired by Apple. At Voysis, he led the development of a suite of technologies that powered voice-enabled experiences for a large number of use cases. Earlier in his career, he held engineering leadership roles at Rapid7, Logentries, and Openet, focusing on cybersecurity, log management, and telecommunications. He also founded Forkstream, a startup specializing in Wi-Fi network selection technology, which was later acquired by Openet.

John holds a PhD in Computer Science from University College Dublin and a degree in Telecommunication Engineering from Dublin City University.