Ronen Schwartz is CEO at K2view.
Discover top fintech news and events!
Subscribe to FinTech Weekly's newsletter
Read by executives at JP Morgan, Coinbase, Blackrock, Klarna and more
The Untold Story Behind the Amazon AI Headlines
When Amazon announced that its AI shopping assistant, Rufus, was now driving massive increases in customer engagement and billions in incremental sales, the reaction was instant: surprise, admiration, and a hint of envy. It was seen as a bold leap forward in how enterprises approach customer experience.
But this wasn’t a triumph of AI models alone. It was made possible by a closed ecosystem. Amazon operates entirely on its own platform, where product, customer, behavioral, and purchase data are unified and controlled. That setup is not a realistic model for most enterprises, especially in financial services. This industry has the highest adoption of AI-powered contact centers, accounting for about a quarter of the global market. Yet its data is still scattered across bank account management, CRM, billing, and support platforms. In environments like these, AI struggles.
The lesson is straightforward: success in customer experience depends less on the brilliance of the model and more on the quality and integrity of the data beneath it. Without a unified, contextual view, AI agents are more likely to disrupt support than to improve it.
When AI Meets a Messy Reality
For most enterprises, the data environment looks nothing like Amazon’s streamlined, vertically integrated platform. Information lives across dozens of systems, each holding pieces of the customer record, duplicated in some places, outdated in others, and rarely in sync.
Dropping AI into that environment creates chaos. Customers receive conflicting or partial responses, trust erodes, and human representatives must step in to restore confidence. What was intended as automation turns into rework, creating heavier burdens on both sides of the conversation.
Think of hiring a skilled service rep but giving them a filing cabinet stuffed with incomplete or mislabeled records. Their talent is wasted because the foundation is broken. The same is true for AI agents: without consistent, accurate, and timely information, they are set up to fail.
What It Really Takes to Scale AI in Customer Experience
Enterprises eager to replicate Amazon’s headlines often zero in on the model itself, fine-tuning prompts, comparing vendors, or chasing the next release. But the deciding factor in long-term success is the data foundation that supports those models.
To make AI agents reliable and enterprise-ready, organizations need three essentials:
- Integration: Customer information spread across dozens of systems must be unified into a single, consistent view.
- Governance and security: Data must be accurate, deduplicated, protected, and compliant with privacy regulations before AI can act on it.
- Real-time context: Agents need the most current information available, not outdated snapshots or static records.
Without these fundamentals, AI quickly unravels, creating errors, compliance risks, and disappointed customers. With them, AI can move beyond pilots to deliver meaningful impact at scale. The lesson is simple but often overlooked: smart agents require smarter data.
From Pilots to Transformation
Across industries, enterprises are experimenting with AI in customer experience, rolling out chatbots, virtual assistants, or generative tools in service workflows. Yet most of these efforts remain stuck in trial mode. A recent MIT report found that nearly 95% of AI projects fail to reach production. Customer experience initiatives are no exception.
The gap between experiment and transformation comes down to the foundation.
Disconnected, poor-quality data undermines support. Clean, unified information enables scale, consistency, and responsible adoption. With the right groundwork, enterprises can finally shift from experiments to production systems that strengthen both customer relationships and business outcomes.
Inspiration and a Warning
The Amazon story is both a milestone and a cautionary tale. It shows what is possible when AI agents are powered by connected, high-quality data, but it also reveals how rare that setup is. Most enterprises cannot simply replicate it. The future of AI in customer experience will not be defined by increasingly sophisticated models alone. It will be shaped by organizations willing to invest in the data foundation that makes those models effective.
