Agentic Commerce Strategy: Closing the AI Spend-Outcome Gap

/ AI, Application Development, Digital Transformation, Machine Learning

The News

Valtech SVP of Strategy and Innovation Megan Carrigan joined ECI Research principal analyst Paul Nashawaty on AppDevANGLE to discuss the state of agentic commerce and AI-driven customer experience. The conversation centered on a persistent tension: enterprise budgets are shifting decisively toward AI, yet consumer adoption of AI-powered commerce experiences is significantly lagging behind the technology’s capabilities. Carrigan argued that most organizations are still operating in conversational AI territory rather than truly autonomous agent-to-agent commerce, and that conflating the two is leading to misaligned investments. Valtech’s advisory stance is to guide clients toward orchestration-first architectures that lean on each platform’s strengths rather than consolidating everything under a single vendor’s roof.

Analyst Take

The Spending Wave Is Real. The ROI Is Not (Yet)

ECI Research’s own data sets up the central tension of this conversation cleanly: according to our 2025 Application Development: Day 1 survey, nearly three in four enterprise IT leaders name AI and machine learning as a top spending priority for the next 12 months. That statistic is no longer a projection; it describes the current state of enterprise capital allocation. Boards have approved the budgets, teams are reorganizing around AI initiatives, and every major platform vendor has repositioned itself as an AI company. The problem Carrigan is describing is what happens after the check clears. Organizations are building faster, models are improving, and infrastructure is scaling, but the customer experience improvements that were supposed to justify the investment aren’t materializing at the same pace. That gap between AI spend and measurable customer outcome is where a lot of organizations are quietly losing ground right now.

The reason, in Carrigan’s framing, is definitional confusion compounding architectural confusion. “Agentic commerce” has become a catch-all term that describes everything from a slightly smarter chatbot to a fully autonomous purchasing agent acting on a consumer’s behalf. Those are fundamentally different design problems, different governance requirements, and different consumer readiness curves. Enterprises that treat them as the same thing end up building for a consumer behavior that doesn’t exist yet. Carrigan’s prescription is to start with conversational experiences that solve a clear job-to-be-done, build trust incrementally, and treat that process as a way to surface content gaps and data quality problems that will matter enormously when true agentic capability becomes viable.

The Platform Consolidation Pitch Is Losing Its Grip

There is a structural dynamic playing out in the enterprise software market that this conversation reflects directly. Every major CRM, CDP, and commerce platform vendor is pitching a consolidated, AI-native end-to-end environment. The reality, as Carrigan puts it, is that most enterprise environments are decades of patched-together systems with data spread across silos. There is no realistic path to consolidating everything under one roof, and organizations that try to do it in response to board pressure on AI timelines are taking on enormous change management risk for uncertain returns.

The orchestration-layer argument Carrigan makes is the more pragmatic alternative: let each platform do what it does best, connect them through agents and governance layers, and use that architecture to extract intelligence that siloed systems can’t produce on their own. This is not a novel architectural concept, but it is gaining urgency because multi-agent systems require coherent orchestration across CRMs, digital asset management platforms, product catalogs, and legacy systems of record simultaneously. Most enterprise architectures were not designed for this kind of cross-system coordination. The organizations that figure out the orchestration layer first will have a durable advantage; the ones that bet on a single vendor’s all-in-one promise will find themselves locked in and under-served when that vendor’s agentic roadmap doesn’t match their actual customer journey.

The Content Layer Is the Underappreciated Dependency

One of the more concrete and immediately actionable insights from this conversation is Carrigan’s point about content strategy. As answer engines and AI agents become the primary discovery layer for consumers, the competitive moat for brand-owned digital properties shifts. Catalog information and comparison data are increasingly being surfaced before a consumer ever reaches a brand’s website. What remains valuable on a brand’s owned experience is depth of context: the brand’s expertise, point of view, and the structured information that helps an agent understand not just what a product is, but when and why a consumer would want it. Carrigan frames this as a new design persona: the agent reader. Content that isn’t structured for agent consumption will underperform in AI-mediated discovery, and that’s a competitive disadvantage that most brands haven’t fully internalized yet.

This connects directly to a broader finding from ECI Research’s 2026 Application Development: DevSecOps + AppSec survey, which identified AI code governance as the #1 priority investment area for enterprise security teams heading into 2026. Governance isn’t just a security problem; it extends to how AI agents are permitted to read, interpret, and act on brand content and customer data. Organizations building agentic commerce experiences without a governance framework for agent behavior are creating liability at the content and data layers, not just the infrastructure layer.

Looking Ahead

The 2025-to-2026 transition referenced at the close of this conversation is one ECI Research tracks closely. Experimentation phases are characterized by small pilots, isolated proof-of-concept projects, and tolerance for ambiguity about ROI. Implementation phases are characterized by the opposite: pressure to demonstrate that the investment translates to customer outcomes, organizational accountability for results, and architectural decisions that are much harder to reverse. Organizations entering that implementation phase without a clear answer to the consumer adoption question Carrigan raises are going to find themselves defending AI spending that hasn’t moved the metrics that matter.

The orchestration-first, governance-forward approach Valtech is advocating will become the dominant consulting and systems integration narrative over the next 18 months, displacing the all-in-one platform consolidation pitch. Vendors that position themselves as orchestration layers rather than destination platforms, and agencies that can operate across the full stack without platform allegiance, are better positioned for this environment than those locked into a single ecosystem. The brands that win agentic commerce are going to be those that started building consumer trust through conversational experiences in 2025, not those that waited for the autonomous agent architecture to mature before engaging. The infrastructure for autonomy is coming; the consumer permission for it has to be earned first.

Author