Today we’re speaking with Dominic Jainy, a distinguished IT professional whose work at the intersection of artificial intelligence, machine learning, and blockchain offers a unique vantage point on our technological future. Our conversation will explore the profound shifts transforming the AI landscape, from the evolution of AI from assistants to autonomous teammates and the critical move toward on-device intelligence for enhanced privacy. We’ll also delve into how software development is being redefined, the groundbreaking partnership between AI and quantum computing, and the new era of regulatory accountability that is reshaping the industry.
We’re seeing a major shift from AI “copilots” to autonomous “teammates” that execute multi-step projects. When creating these “Digital Assembly Lines,” what are the biggest challenges in orchestrating different specialized agents, and how do you ensure they collaborate effectively without human intervention?
That’s the core of the transition from generative to agentic AI. The real challenge isn’t just creating a single smart agent; it’s getting a whole team of them to function as a cohesive unit. Imagine a marketing agent drafts a campaign. It then has to seamlessly hand that off to a legal agent, which must understand the context, review it for compliance, and pass it to a third agent for scheduling. The primary hurdle is creating a common language and reasoning framework so they can plan and execute these multi-step projects without stumbling. It requires a sophisticated orchestration layer that manages not just tasks, but the flow of context and intent across different specialized domains, truly moving beyond the simple “chat-and-response” model.
For industries like finance and healthcare, on-device AI is becoming the standard for ensuring privacy. What are the primary trade-offs a company must weigh when moving from large cloud-based models to local AI, and could you detail the steps for implementing this transition?
The fundamental trade-off has always been power versus privacy. Massive cloud models offer unparalleled computational strength, but they require you to send sensitive data off-premise. Local AI, powered by new specialized AI chips and Small Language Models, puts privacy first by keeping everything on the device. For a company in finance, the first step is investing in hardware with capable NPUs. Next, you must re-architect your software to leverage these local capabilities, often using smaller, highly optimized models. Finally, the focus shifts to creating applications that benefit from zero-latency interactions, like instant augmented reality overlays or real-time voice translation, which simply aren’t feasible when you’re waiting for a round trip to the cloud. It’s a strategic decision to prioritize security and responsiveness over raw scale.
As development moves from writing code to expressing intent in natural language, what are the most critical new skills for software professionals? Please describe how this paradigm shift changes the day-to-day workflow and team structure in a modern technology department.
The ground has completely shifted under our feet. The most critical skills are no longer about mastering the syntax of a specific programming language, but about System Architecture and Orchestration. A developer’s value now comes from their ability to clearly express a desired outcome and design a system where AI agents can build and manage the underlying deterministic logic. In a typical day, a developer might spend less time debugging lines of code and more time refining prompts, defining agent roles, and overseeing the system’s architecture. This changes team structures, too; we see more cross-functional pods where strategists, ethicists, and architects work together to guide AI systems, rather than siloed teams of coders.
Self-healing enterprise software that autonomously patches bugs is now a reality. Could you walk us through the process of how an AI agent identifies a software flaw and deploys a fix? Please share the key metrics a CTO should track to measure its impact.
It’s a fascinating process that feels like something out of science fiction. When an error occurs, an AI monitoring agent first identifies the anomalous behavior. Instead of just logging an error code, it analyzes the context—what the user was doing, the state of the system—to diagnose the root cause. It then tasks a specialized coding agent to generate a potential patch. This patch is tested in a sandboxed environment, and if successful, deployed autonomously. The most crucial metric for a CTO is the reduction in maintenance overhead, which we’re seeing can be up to 40%. Other key indicators are mean time to resolution (MTTR) for bugs, which is plummeting, and a decrease in customer-reported incidents, as the system often fixes flaws before a human even notices.
The partnership between AI and quantum computing is accelerating discovery in fields like pharmaceuticals. Besides drug development, which industries are most poised to benefit from this hybrid approach, and what are the first practical steps for a business to begin exploring these capabilities?
While pharma gets the headlines for its work in molecular discovery, the finance and logistics industries are right behind. In finance, hybrid systems can model incredibly complex market scenarios that are impossible for classical computers, optimizing investment strategies. For logistics, they can solve routing problems on a global scale with unprecedented efficiency. A business looking to start should first identify a problem that involves simulating the unknown, not just analyzing past data. The initial step isn’t to buy a quantum computer, but to partner with cloud providers offering quantum simulation services. This allows them to build quantum-ready algorithms and experiment with this new paradigm without a massive upfront investment, preparing them for when the hardware becomes more accessible.
With regulations like the EU AI Act now in full effect, companies must provide “Explainability Reports” for high-risk decisions. What does a comprehensive report involve, and how can organizations build systems that are not only compliant but also genuinely transparent and trustworthy?
A comprehensive “Explainability Report” is far more than a data sheet. For a high-risk decision, like a loan denial, it must detail the entire decision-making chain. This includes the source and provenance of the training data, the specific features that most influenced the outcome, and a clear, human-readable justification for the AI’s conclusion. To build these systems, organizations must move beyond “black box” models. This involves embedding transparency from the ground up, using inherently interpretable models where possible, and conducting rigorous AI supply chain audits. It’s about creating a culture where proving fairness and providing clear reasoning is just as important as the accuracy of the model itself.
The human role is evolving into a “Human Supervisor” focused on strategy and ethical oversight. What training and new organizational structures are needed to support this transition, and how do you measure the performance of an employee whose primary job is to direct AI agents?
This is a complete reimagining of professional development. We need training programs focused on strategic thinking, ethics in AI, and prompt engineering—the art of asking the right questions. Organizationally, this means flatter structures where “Human Supervisors” act as conductors for teams of AI agents, much like a project manager. Performance is no longer measured by output, like lines of code written or reports filed. Instead, we measure the effectiveness and efficiency of the AI systems they direct. Key performance indicators become the overall business outcomes of their AI team, the ethical compliance of their agent’s actions, and their ability to innovate by creatively combining different AI capabilities to solve complex problems.
What is your forecast for the next major leap in artificial intelligence beyond 2026?
Looking beyond 2026, I believe the next monumental leap will be the emergence of AI systems capable of generalized problem-solving across completely unrelated domains. Today’s agentic AI is powerful but specialized. The next step is an AI that can take learnings from optimizing a supply chain and apply that abstract knowledge to, say, managing a city’s power grid or discovering new principles in physics. This isn’t just about being better at specific tasks; it’s about achieving a level of abstract reasoning that allows AI to become a true partner in scientific discovery and societal-level problem-solving, moving from a director of tasks to a genuine collaborator in human progress.