Establishing a Strong AI Foundation for CX

Author: Beth Schultz, VP of Research and Principal Analyst

How the most successful companies differentiate on AI governance, data architecture, and model agility  

As artificial intelligence (AI) increasingly drives customer experience (CX) transformation, IT and CX leaders face critical decisions regarding architecture, model selection, and data governance. After all, merely implementing AI doesn’t guarantee return on investment (ROI).

In our AI Technology Foundation & Strategy: 2026-27 research study of 759 companies globally, Metrigy analyzed measured business outcomes to isolate participating organizations that have achieved above-average improvements across several key metrics from their AI initiatives: revenue growth, cost decrease, cost avoidance, efficiency boost, and improvements in customer satisfaction (CSAT) and employee satisfaction (ESAT) ratings. In examining the technological footprints of these top-performing organizations—our research Success Group—we see clear patterns emerge specifically regarding the use of AI for CX. Among many other ways, the Success Group is gaining value from AI for CX by:

• Redesigning how work gets done, yielding a bigger drop in future costs and higher customer satisfaction compared to just automating existing workflows
• Using customer interaction data to directly compare the performance of AI agents against human agents, for continuous coaching and improvement opportunities
• Implementing AI orchestration overlays directly on agent desktops, allowing integration of all agent applications into a single interface and improving agent efficiency by decreasing the need to toggle between applications

The ability to optimize AI-for-CX initiatives requires a strong technology foundation and strategy. In this report, we delve into how successful companies are optimizing their AI technology foundations and strategies, with a focus on their use of:

• Rigorous governance
• Advanced data architectures
• Strategic model agility

Table of Contents

• Introduction: Establishing a Baseline for AI-for-CX Success
• Elevating Governance and Risk Management
  • From the Theoretical to the Operational
  • Human in the Loop
• Activating Unstructured Data and RAG Architectures
  • Layered Data Architecture
  • Retrieval-Augmented Generation & Unstructured Data
• Embracing Model Agility and SLMs
  • Frontier Models & Data Exchange
  • From Large to Small Models
  • Exit Strategy for Moving from One Model to Another
• Conclusion & Recommendations
• Working With Metrigy