How Agentic AI Turns Your Biggest Tech Problem into Competitive Advantage

Agentic AI has officially dethroned generative AI as the darling of the technology hype cycle. Where last year’s Boardrooms buzzed with ChatGPT demonstrations and prompt engineering workshops, today’s conversations fixate on autonomous agents that can plan, execute, and adapt without human intervention. Gartner predicts1 that by 2029, agentic AI will autonomously resolve 80% of common customer service issues without human intervention, leading to a 30% reduction in operational costs. However, there is a troubling paradox at play: organisations racing to capture this transformative potential are attempting to deploy these revolutionary systems atop technological foundations that predate the iPhone.
This paradox reminds me of the cloud transformation journeys I’ve witnessed at AWS. Organisations began with pragmatic ’lift and shift’ migrations for speed and business continuity. They then moved to modernisation — adopting managed services, adding auto-scaling, perhaps taking their first tentative steps with serverless architectures. But this modernisation was often just tinkering around the edges, making incremental improvements whilst leaving core architectural decisions untouched. Now, with agentic AI, we’ve reached the third step that many avoided: complete refactoring to truly cloud-native architectures. The autonomous systems of today can’t function effectively with half-measures. They require what I’ve been advocating for years: real-time data access, loosely-coupled designs (where changes in one area don’t break others), event-driven architectures (that respond instantly to changes), and API-first approaches. In many ways, agentic AI is forcing the architectural transformation that organisations have been deferring for over a decade.
This mismatch between agentic AI’s potential and legacy architectures presents Boards with a choice that will define their competitive position for years to come. They can deploy agentic AI within existing constraints, accepting that these systems will operate at a fraction of their potential; autonomous in theory but constantly hamstrung by technology and processes that are beyond their use-by date. Or they can recognise this moment for what it is: an unprecedented opportunity to use these autonomous systems as the catalyst for retiring decades of technical debt. What makes this particularly compelling is that the very characteristics that make agentic AI transformative — its ability to understand complex systems, identify inefficiencies, and autonomously implement solutions — also make it uniquely capable of leading this architectural evolution. The organisations that grasp this opportunity won’t just be implementing new technology at full strength; they’ll be clearing the path for the next wave of AI and quantum-related technologies that will define the next era of business competition.
Understanding Agentic AI
For those new to the concept, agentic AI goes beyond traditional artificial intelligence that merely responds to queries or analyses data to predict an outcome. These systems possess genuine autonomy - they understand goals, devise strategies, execute plans, and adapt based on outcomes, much like a trusted colleague handling complex tasks independently.
Consider supply chain management: a traditional AI might analyse your inventory levels and flag potential shortages. An agentic AI system goes further, it detects the shortage risk, checks multiple supplier inventories, compares pricing and delivery times, negotiates terms within pre-set parameters, places orders, arranges shipping, updates your financial systems, and notifies relevant teams, all without human intervention. It’s this ability to act across multiple systems and make decisions autonomously that makes legacy architectures so problematic. When your procurement system only updates overnight, when supplier data sits in separate silos, or when approval workflows require manual sign-offs, these agents hit wall after wall. They can’t navigate effectively when critical data is locked away or when processes assume human-speed decision-making.
The Real Barriers to Agentic AI Adoption
Through the lens of the AI Stages of Adoption (AISA), I see organisations stuck in the Adopting and Optimising stages, unable to progress to Transforming and Scaling. Whilst research shows 53% of leaders cite data privacy concerns and 76% blame talent shortages2, these visible barriers mask a deeper truth. They’re symptoms of organisations trying to force transformative technology into structures never designed to support it. When 75% of enterprise workloads remain on premises3 — many built before modern API standards — every other challenge amplifies. How can you ensure data privacy when lineage spans decades? How can you trust AI decisions from systems with layers of undocumented logic?
The human and organisational barriers prove equally interconnected with technical foundations. McKinsey identifies trust, governance, and clarity on human-agent workflows as core blockers4, but these intensify when agents must navigate approval hierarchies designed for quarterly reviews, not microsecond decisions. Without unified architecture, even the best governance frameworks struggle to provide meaningful control whilst avoiding the “uncontrolled agent sprawl” that turns automation into chaos.
This interconnected nature explains why Gartner predicts 40% of agentic AI projects will face cancellation by 20275 — not from technology failure but foundational unreadiness. C-suite confidence has already dropped 11 percentage points year-over-year, down to 58% according to Akkodis6, with executives recognising their organisations aren’t architected for AI’s promise. The complete AI adoption framework — AISA, the Five Pillars, and Well-Advised — reveal how these barriers compound: Technical Infrastructure gaps impact governance capabilities, poor cultural readiness stems from training teams on hostile systems, and value measurement fails when data remains siloed. These aren’t separate problems, they’re manifestations of the same architectural constraints.
From Debt Retirement to Revolutionary Enablement
But here’s where the narrative shifts from challenge to opportunity. The very characteristics that make agentic AI struggle in legacy environments also make it uniquely capable of leading your technical transformation at unprecedented speed and scale.
Consider the economics: McKinsey estimates7 that technical debt consumes 40% of IT budgets in large enterprises. Traditional modernisation projects take years, cost millions, and often fail when key personnel leave. Agentic AI fundamentally changes this equation. These systems work 24/7, never forget what they’ve learned, and can modernise entire architectures whilst your business continues to operate.
The business case is compelling. According to McKinsey4 a large bank modernising its legacy core system — comprising 400 pieces of software and budgeted at over $600 million — deployed squads of AI agents to autonomously document applications, write and review code, integrate features, and test outputs. What traditionally involved slow, manual processes across siloed teams was accelerated, achieving a more than 50% reduction in time and effort for early adopter teams. The agents didn’t just refactor code, they enabled a more efficient architecture that shifted human roles to oversight, unlocking faster delivery of new features.
This isn’t science fiction. AWS Transform3 already uses agentic AI to modernise mainframe applications 50% faster than traditional methods, reducing costs by up to 40%. But the real opportunity lies in applying these capabilities strategically across your entire technology estate. When agents rebuild your systems, they don’t just fix old problems—they create architectures that become competitive weapons. The real-time, event-driven systems they build enable instant customer response whilst establishing data governance frameworks that transform compliance from burden to differentiator. Most critically, they embed the flexibility your organisation needs to adopt whatever technologies emerge next, turning today’s transformation into tomorrow’s platform for innovation.
This transformation directly accelerates progression through the AISA stages. Organisations stuck in Optimising due to architectural constraints suddenly find the path to Transforming cleared. When agents eliminate the technical barriers that kept you from reimagining business models, when they build the infrastructure that enables ecosystem-wide integration, you naturally progress toward Scaling. The same debt retirement that reduced time and effort by more than 50% for early adopter teams4 also unlocks the architectural capabilities required for advanced AISA stages.
The strategic implication is clear: organisations using agentic AI to retire technical debt aren’t just reducing costs, they are building the foundation for capabilities their competitors can’t match. When your systems can process data in milliseconds or less rather than overnight batches, when new products launch in days rather than months, when compliance updates happen automatically rather than through manual reviews, you’re not just more efficient. You’re playing a different game entirely.
Debt Retirement Cascades into Innovation
The strategic significance of using agentic AI for technical transformation extends far beyond operational efficiency. By clearing today’s architectural constraints, organisations position themselves to adopt technologies that will define competitive advantage in the 2030s — capabilities that are impossible to implement on legacy foundations.
Quantum-AI hybrids hold strong potential to transform drug discovery and financial modelling, but they require high-quality, real-time data flows that many organisations still lack. Pharmaceutical companies attempting quantum-enhanced simulations on overnight-batch systems often face severe inefficiencies, as the technology relies on seamless integration. Legacy architectures pose three key barriers: batch processing hinders timely handoffs between quantum and classical computing, data silos impede holistic analysis, and inflexible structures struggle to meet quantum’s specialised demands. However, organisations leveraging agentic AI to modernise with real-time processing and unified data models can effectively overcome these hurdles, gaining a competitive edge in adopting quantum technologies.
The same principle applies to neurosymbolic AI—systems that integrate neural networks’ pattern recognition with symbolic reasoning’s interpretability. Financial services firms show growing interest in these systems for their potential to explain complex decisions, which is essential for regulatory compliance. Neurosymbolic AI combines AI’s pattern-spotting capabilities with auditable rule-based reasoning. To perform effectively, it requires clean, interconnected enterprise data and well-defined digital models of business operations, elements often limited by legacy silos. These needs align closely with the modern architectures enabled by agentic AI transformations. When agents modernise systems, they go beyond data migration to foster semantic understanding and structural adaptability, better supporting evolving AI paradigms.
Perhaps most tangibly, embodied AI — autonomous systems that interact with the physical world — is projected to significantly transform manufacturing, logistics, and retail by 2030. Embodied AI is rapidly advancing toward broader adoption in factories and logistics. It demands low-latency responsiveness and seamless integration between shop-floor machines and enterprise systems—capabilities often lacking in outdated, batch-oriented architectures. Manufacturers reliant on overnight processes and disconnected operational technology face major challenges in deploying autonomous robots effectively. However, organisations leveraging agentic AI for modernisation are creating the event-driven, loosely-coupled architectures essential for embodied AI’s future demands.
This cascade effect creates compound returns on transformation investments. The board that approves agentic modernisation isn’t just reducing today’s IT costs, they’re building optionality for technologies that don’t fully exist yet. Through the lens of the Five Pillars capabilities areas, you’re simultaneously strengthening Technical Infrastructure, enabling better Governance & Accountability through cleaner systems, and creating the foundation for advanced Value Realisation & Lifecycle Management. When evaluating these initiatives, a comprehensive business case must capture this multi-horizon value across all Well-Advised dimensions: immediate Operational Excellence, medium-term Customer Value through superior architecture, and long-term Innovation positioning. The organisations that grasp this cascading value won’t find themselves scrambling to adopt each new technology wave — they’ll be ready to lead it.
Questions Every Board Should Be Asking
As you consider your organisation’s approach to agentic AI, several strategic questions deserve your attention:
Is your technical debt creating a ceiling on your AI ambitions? If your agents can access only fragments of your data, if they must wait for overnight batch processes, or if they’re constrained by manual approval gates, you’re not just limiting today’s performance. You’re potentially excluding your organisation from the quantum-enhanced, neurosymbolic future of business competition. What would be possible if your agents had unfettered access to real-time, comprehensive data?
Are your agents building tomorrow’s platform or tomorrow’s prison? When agentic AI systems redesign your processes and architectures, they encode assumptions about how business should work. Will those assumptions enable flexibility and evolution, or will they lock in today’s constraints? The difference lies in whether you’re using agents to optimise existing processes or fundamentally reimagine them for an AI-native world.
When embodied AI arrives at your industry’s door, will your infrastructure speak its language? Within five years, autonomous systems will be standard in manufacturing, logistics, and retail. These systems demand sub-millisecond response times, edge-to-cloud orchestration, and seamless physical-digital integration. If you’re still running batch processes and disconnected systems, you won’t just be behind—you’ll be incompatible with the future of your industry.
How will you govern at the speed of autonomous decision-making? As I’ve written previously, Boards are moving from overseeing hundreds of decisions per day to millions per second - and must be confident that each is transparent, explainable, and correct. Traditional governance assumes human-speed reviews and quarterly checkpoints, but agents operate at an entirely different velocity. Can your governance frameworks provide meaningful oversight without becoming bottlenecks? Where should human judgment remain paramount, and where must you trust algorithmic decision-making? The answer determines whether AI becomes your most powerful asset or your greatest ungoverned risk.
What happens when your competitors’ agents start collaborating? The future isn’t just about internal automation — it’s about agent-to-agent commerce. When your suppliers’ agents can negotiate directly with your customers’ agents, when entire value chains become autonomous ecosystems, will your systems be able to participate? Or will you be forced to maintain costly human intermediaries whilst others transact at the speed of light?
The Window of Strategic Opportunity
We stand at a rare inflection point where a technology’s greatest challenge — integrating with legacy systems — also represents its greatest opportunity. Agentic AI doesn’t just struggle with technical debt; it can systematically eliminate it whilst building foundations for capabilities we can barely imagine today.
The organisations that recognise this moment won’t approach agentic AI as another IT project to manage. They’ll see it as the key to unlocking decades of trapped value whilst positioning for technological advantages their competitors can’t match. They’ll use agents not just to automate today’s processes but to build tomorrow’s platforms.
The questions I’m fielding have evolved remarkably over the past two years. Where once executives asked “Should we be looking at AI?” they now ask “How do we deploy agents effectively?” But the most strategic leaders are asking a different question entirely: “How do we use AI to build a company that thrives in 2035?”
That question acknowledges a fundamental truth: in the age of exponential technological change, your ability to adopt tomorrow’s innovations depends on the architectural decisions you make today. Every month spent operating with technical debt isn’t just a month of inefficiency — it’s a month further behind in the race to build AI-native capabilities. Fast followers will become permanent followers.
The window for this transformational approach won’t remain open indefinitely. As agentic AI becomes standard, the advantage will shift from those who adopt it to those who use it most strategically. The question isn’t whether to embrace agentic AI—it’s whether you’ll use this moment to clear the path for every innovation that follows.
Your next Board meeting presents a clear imperative: mandate agentic AI pilots explicitly focused on technical debt retirement within the next 12 months. Start with one critical system — perhaps customer data consolidation or supply chain integration — where legacy constraints most limit your ambitions. Set clear metrics: not just cost reduction, but architectural readiness for future technologies. Without this deliberate action, you risk architectural lock-in that will exclude you from the AI-defined future. The time for strategic committees and feasibility studies has passed. The time for architecting your competitive advantage is now.
Let's Continue the Conversation
Thank you for reading my thoughts on agentic AI and technical debt retirement. If you'd like to discuss how these concepts apply to your organisation or share your experiences with emerging AI technologies, I welcome the opportunity to exchange ideas.
https://www.gartner.com/en/newsroom/press-releases/2025-03-05-gartner-predicts-agentic-ai-will-autonomously-resolve-80-percent-of-common-customer-service-issues-without-human-intervention-by-20290 ↩︎
https://www.cloudera.com/blog/business/ready-to-scale-tackling-the-top-challenges-of-agentic-ai-adoption.html ↩︎
https://aws.amazon.com/blogs/migration-and-modernization/aws-transform-generally-available/ ↩︎ ↩︎
https://www.mckinsey.com/capabilities/quantumblack/our-insights/seizing-the-agentic-ai-advantage ↩︎ ↩︎ ↩︎
https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027 ↩︎
https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/breaking-technical-debts-vicious-cycle-to-modernize-your-business ↩︎
About the Author
Mario Thomas is a transformational business leader with nearly three decades of experience driving operational excellence and revenue growth across global enterprises. As Head of Global Training and Press Spokesperson at Amazon Web Services (AWS), he leads worldwide enablement delivery and operations for one of technology's largest sales forces during a pivotal era of AI innovation. A Chartered Director and Fellow of the Institute of Directors, and an alumnus of the London School of Economics, Mario partners with Boards and C-suite leaders to deliver measurable business outcomes through strategic transformation. His frameworks and methodologies have generated over two-billion dollars in enterprise value through the effective adoption of AI, data, and cloud technologies.