
Source: Ivan S via Pexels.
The organizational push toward agentic AI is redefining customer experience strategy, with an overwhelming 73% of organizations agreeing that the ability of AI to act automatically creates a real competitive advantage. This strategic urgency is tangible, as a study conducted by 451 Research by S&P Global shows 70% of organizations are actively reevaluating their current technology vendors due to the arrival of new AI capabilities. Yet the route to automation is neither unified nor straightforward. Organizations are divided across build-and-buy strategies, constrained by integration debt and grappling with a pricing model transition that is beginning to redistribute risk between buyers and vendors.
The Take
The prevailing narrative frames agentic customer experience (CX) as an automation race, but survey data shows a more constrained, pragmatic picture. Organizations are not pursuing full autonomy at all costs; they are augmenting human agents first, with increasing agent productivity (24%) and providing real-time guidance (22%) ranking ahead of full containment as immediate AI priorities. The build-vs-buy market is fragmented—no single deployment strategy commands more than a 28% share—which means integration complexity, not AI capability, is the primary execution risk. High vendor loyalty (73%) coexists with a strong appetite for new technology (63%), indicating a strategic shift toward augmenting core platforms with new AI vendors rather than replacing them. This dynamic signals a market demanding both stability and innovation, where winning vendors will be those that offer flexible deployment, seamless integration and transparent, outcome-based pricing.
Summary of findings
Customer satisfaction is the primary business driver, especially in business-to-consumer (B2C) organizations. Improving customer satisfaction is the top motivation for investing in AI-driven voice channels (34%), outpacing operational cost reduction (20%). This focus is more pronounced in organizations with a B2C component, where 40% cite customer satisfaction as the main driver, compared with just 26% of pure-play business-to-business firms. The higher volume and immediacy of consumer interactions appear to make agentic deployment a more urgent priority in B2C contexts.
Conversational voice AI has reached a meaningful adoption threshold. Voice AI is moving from testing to live deployment. Survey data shows the market at an inflection point, with organizations split evenly between production (29%) and pilot/proof-of-concept stages (29%). This suggests the market is scaling and operationalizing rather than being fully standardized.
The build versus buy market remains fragmented. There is no single dominant strategy for acquiring agentic AI capabilities. Building custom solutions on hyperscaler platforms (AWS, Google, Microsoft) is the most common approach (28%), followed closely by using the native AI capabilities of a core contact-center-as-a-service/contact center platform (25%) or a primary CRM vendor (22%). In contrast, specialized conversational AI vendors (Sierra, Decagon, Maven) represent a meaningful but somewhat smaller primary route (18%). This indicates that the decision is highly dependent on an organization’s existing infrastructure, in-house technical skills and desired level of control over the AI models.
Outcome-based pricing is the new buyer demand. As organizations move from experimentation to accountability, pricing models are under scrutiny. While a base-fee-plus-usage model is the most common today (25%), when asked to choose one model, buyers prefer outcome-based pricing (where payment is tied to verifiable results, such as issue resolution) (18%). This shift is driven by significant budget concerns: 31% of organizations view unpredictable consumption costs as a major risk, and 30% cite a lack of cost visibility as a direct barrier to adoption. This signals a clear demand for vendors to share risk and align their commercial models with measurable CX performance.
Agent augmentation is the immediate priority. The most critical AI capabilities sought by organizations are those that empower human agents rather than replace them entirely. Increasing human agent productivity and performance (24%) and providing real-time guidance and data to service representatives (22%) are the top two priorities. This reflects a strategic focus on achieving near-term return on investment (ROI) by making the existing workforce more effective before moving to fully autonomous systems.
The proportion of respondents reporting AI agents “in production” overstates true autonomy by roughly 2.5 to 3 times. Across the 18 surveyed agentic use cases, an average of 29% of organizations report having agents in production. However, this figure includes truly autonomous agents and those with a human in the loop. Isolating genuine autonomy tells a different story:
A separate question shows that among agents taking actions, only about 40% operate without a human approval step. Applying that ratio, truly autonomous production collapses to roughly 12% per use case, and no category — marketing, commerce or service — exceeds approximately 15%.
That gap is not a rounding artifact; it reflects the human-in-the-loop layer that a simple in-production count conceals. For vendors and buyers alike, the “deployed” figure marks where agents are live, not where they run unattended — a critical distinction for ROI models, staffing plans and governance design.
Most deployed agentic AI still runs with a human in the loop. Asked what level of autonomy AI has today, only about 19%-22% of organizations report AI that acts automatically, with no human approval needed across customer experience activities — the genuinely agentic tier. A larger 26% to 33% sit one step back, where AI acts but a human approves at key steps, and roughly 30% remains advisory, with AI suggesting and a human deciding. In practice, agentic CX is less an autonomy race than a supervised-augmentation phase: Organizations are expanding what agents can do while retaining human accountability at decision points. The pattern aligns with stated buyer priorities, where increasing agent productivity (24%) and real-time guidance (22%) outrank full containment. Full autonomy is the trajectory — but human oversight remains the default operating mode.
Agentic AI vendor loyalty is strong, yet value is crucial. Most organizations (73%) report they are very likely to continue with their primary agentic AI vendor. The top reasons for this loyalty are reliable platform performance (34%) and good value for money (33%). For the small subset of organizations likely to switch, poor value for money (45%) and the desire to pursue a multi-vendor strategy (48%) are the leading drivers. This suggests that while performance ensures stability, cost-effectiveness is the key to long-term retention.
AI capabilities are prompting a broader reevaluation of CX vendors. The rapid evolution of AI is disrupting established vendor relationships. A significant 70% of organizations report reevaluating their CX technology vendors because of new AI capabilities. This suggests a market in flux, where incumbent vendors can no longer rely on inertia. Instead, they must continually innovate and demonstrate a clear AI road map to retain customers who are actively seeking the most advanced capabilities.
Connected data is the new currency for AI success. AI is amplifying the value of a unified data strategy. A commanding 76% of organizations agree that connected customer data is now more important, specifically because of their AI initiatives. This highlights a critical dependency: the effectiveness of any agentic AI system is directly tied to the quality and accessibility of the underlying customer data. It signals that successful AI deployment is inseparable from a modern, cohesive data infrastructure, creating new urgency for data integration projects.
Agentic AI adoption is highest in customer service and commerce. When examining specific use cases, agentic AI is most mature in structured, high-impact areas. In customer service, intelligent routing and triage lead, with 34% of organizations having it in production. In commerce, checkout assistance (31%) and subscription/renewal management (30%) show the highest production usage. These use cases reflect a strategic focus on deploying autonomous agents to streamline critical points in the customer journey, where efficiency and accuracy directly impact revenue and retention.
Appetite for new vendors reveals a multi-vendor reality. Despite strong loyalty to primary vendors, there is a strong appetite for new technology, with 63% of organizations stating they are very likely to purchase from a new agentic AI vendor in the next purchasing cycle. This openness is particularly high in the finance (75%) and hospitality (74%) sectors. Interestingly, tech laggards (73%) are more likely than tech leaders (59%) to seek new vendors, suggesting they may be trying to leapfrog capabilities by acquiring innovative, stand-alone solutions rather than expanding their existing, more integrated stacks.
Want insights on AI trends delivered to your inbox? Join the 451 Alliance.
This content may be AI-assisted and is composed, reviewed, edited and approved by S&P Global in accordance with our Terms of Service.
The author(s) used an AI platform in the production of the report this blog post was based on. It was subsequently peer-reviewed, fact-checked and edited before publication. S&P Global uses generative AI to create content in accordance with our Terms of Use.

