AI and ML
Agentic AI and the new dynamics of retail
Executive overview
Retail is undergoing a fundamental transformation in how consumers discover and purchase products. The rise of autonomous AI agents is ushering in an era where shoppers are increasingly using agents to discover the products they like, evaluate alternatives, and make purchasing decisions, often with greater speed and precision than any human. This change reshapes the fundamentals of customer engagement, business operations and competitive advantage.
The numbers reveal a rapidly closing window of opportunity. AI agents will drive 25% of e-commerce by 2030 (source). Consumer behavior is already shifting to match this technological leap. Currently 23% of Gen Z and 27% of Millennials trust AI product recommendations more than human ones (source).
For today’s retailers, this wave brings both significant challenges and incredible opportunities. On one hand, traditional digital strategies like static user interfaces and SEO are quickly losing effectiveness in an environment shaped by real-time, AI-mediated decisions. On the other hand, businesses willing to embrace an agentic approach can unlock new value by building intelligent ecosystems, owning relationships across both human and AI-driven channels and differentiating themselves in a crowded market.
Success will depend on your readiness to adapt. The ability to transform legacy systems, create a robust data foundation and build trustworthy, secure machine-facing infrastructure will determine your relevance to tomorrow’s shopper. This white paper lays out the business case and urgent imperative for action, explores the landscape of emerging opportunities and risks, and provides a practical roadmap to lead your brand through the shift to agentic commerce.
We see a stark disconnect in the market. While over 50% of consumers plan to use AI shopping assistants, only 11% of retailers are prepared to scale AI solutions because of fragmented data (source). This gap between consumer adoption and retailer readiness represents a significant opportunity for forward-looking organizations. By investing in a unified, agent-ready infrastructure, you can position your brand at the center of this emerging agentic economy where AI mediates purchase decisions rather than direct human interaction.
The evolution of agentic commerce
The transition from traditional search-driven commerce to AI-mediated ecosystems is happening at a pace that surpasses typical retail technology cycles. Global agentic commerce is projected to reach $3 trillion to $5 trillion by 2030 (source), with year-over-year growth in AI-driven retail traffic hitting 4,700%(source). While many organizations recognize that existing strategies are losing effectiveness, most are still in early experimentation phases.
Current initiatives, such as deploying chatbots, enabling natural-language search, or generating AI-assisted product descriptions, represent incremental improvements. They enhance user experience but do not fundamentally change how commerce operates.
A fully realized agentic ecosystem goes much further. It requires systems where AI agents can interpret product catalogs correctly, reliably access real-time inventory, understand nuanced customer intent and independently complete transactions without human intervention. (This is evidenced by agents moving from concept to operations, supporting production workflows like operations, security and marketing.) This level of capability depends on deep integration across data, infrastructure and business logic.
At present, very few retailers have achieved this level of maturity. Most remain constrained by legacy systems that require manual oversight and fragmented workflows. This disconnect between basic AI adoption and true autonomy represents a significant opportunity for forward-looking organizations to differentiate themselves over the next decade.
Key challenges to adoption
Despite the clear opportunity, the path to agentic commerce is complex. Organizations must address several foundational challenges:
Architectural Friction
Many retailers operate on layers of legacy systems that have been patched together over time. These architectures make it difficult to expose clean, scalable APIs or enable real-time data exchange. As a result, critical information remains locked in silos, limiting accessibility for AI systems.
Data readiness
AI agents rely on structured, consistent and semantically meaningful data to function effectively. 58% of retailers report fragmented data that hinders agent effectiveness. In practice, many product catalogs contain inconsistencies, missing attributes and inaccurate inventory signals. Without significant data normalization and enrichment for intent, AI agents cannot reliably interpret or recommend products. Hence a trusted data foundation is a core prerequisite, including governance, semantic layers and real-time pipelines for AI-ready data.
Security and trust
Agent-driven commerce introduces new risks and responsibilities. Autonomous transactions require robust authentication mechanisms, secure payment infrastructures and advanced fraud detection. At the same time, consumers and businesses must feel confident allowing AI systems to act on their behalf, which raises the bar for transparency and trust. This requires a new focus on security for the AI era, including runtime protection, exposure management and secure-by-design cloud operations.
The owned channel: Risk and opportunity
For more than a decade, digital retail strategy has centered on driving traffic through SEO, optimizing for mobile experiences and guiding human users through carefully designed interfaces. This model assumes that consumers are actively browsing and making decisions themselves.
That assumption no longer holds. Today 88% of Gen Z and Millennials expect generative AI to transform how they shop. They are moving rapidly from search-and-click to delegate-and-verify. As AI agents take over discovery and evaluation, brands are increasingly represented indirectly through data consumed by external platforms.
This creates a significant risk. When third-party AI systems mediate the customer relationship, they often reduce brand differentiation to a limited set of attributes such as price, availability and ratings. The richness of brand storytelling and experience can be lost, eroding long-term equity.
However, this shift also presents an opportunity. Retailers can reclaim control by transforming their owned channels into agentic, omnichannel ecosystems. These environments function not just as storefronts, but as intelligent interfaces capable of understanding complex goals, curating holistic solutions, self-healing and supporting both human users and AI agents.
A critical enabler of this transformation is a unified infrastructure. Rather than building separate systems for internal experiences and external integrations, retailers should establish a single data and service layer that supports both. With the right governance and guardrails, organizations can control how much context is shared externally while maintaining consistency across all touchpoints.
Retailer readiness
Across the industry, organizations are navigating the transition to autonomous commerce at different velocities:
Architects of the future
These organizations are proactively modernizing their infrastructure, dismantling legacy constraints and investing in the semantic data and API-first architectures required for machine navigation. They are strategically positioned to not only capture demand but to define the rules of the new economy.
Strategic adapters
These companies are successfully experimenting with agentic technologies but are still working toward a cohesive, enterprise-wide infrastructure. While they recognize the shift, their current incremental approach requires a pivot toward a unified control plane to compete at scale.
Legacy-centric organizations
These retailers remain optimized for traditional "search-and-click" models centered on static content. As AI agents become the primary interface for discovery, these organizations are at a critical crossroads. They need to modernize foundational data to remain visible to the autonomous shopper.
A practical roadmap
Successfully transitioning to agentic commerce requires a phased, strategic approach:
1. Improve visibility
Retailers must ensure their data is accessible and interpretable by AI systems. This includes restructuring websites and backend systems so that product, pricing and inventory data can be exposed through machine-readable formats and real-time APIs.
2. Enable conversational commerce
Static navigation and filtering should be replaced with systems that can interpret intent and respond dynamically. These systems can individualize recommendations, suggest complementary products and even negotiate pricing or bundles based on context.
3. Support autonomous transactions
Organizations need to implement secure frameworks that allow AI agents to complete purchases independently. This involves advanced authentication, authorization protocols and seamless payment integrations designed for machine-to-machine interactions.
4. Automate post-purchase experiences
The role of AI should extend beyond the point of sale. Returns, refunds, customer support and service inquiries can all be handled by agents, reducing operational overhead while improving responsiveness.
5. Redefine the value chain
As agentic ecosystems mature, new business models will emerge. Retailers can explore outcome-based pricing, new partnerships, subscription models driven by AI optimization and deep integrations within multi-agent environments that coordinate across multiple providers.
Enabling technology
To survive the shift from human browsing to agentic delegation, a retailer’s technology stack must evolve from a collection of "human-readable" web pages into a Machine-Navigable Brand Ecosystem. This transformation is not a singular software update but a structural reconfiguration across two distinct layers: the Foundational Infrastructure of Trust and the Agentic Interaction Layer.
The foundational blocks: Building the nervous system
Before an agent can act, it must reason over your brand. This requires a shift in how you expose data and logic:
Semantic data and intelligence (the brain): Traditional flat SKUs are an artifact of the UI-first era. You need to transform product catalogs into Semantic Knowledge Graphs using vectorized structures. This ensures your brand is not just indexed, but becomes the "best answer" for complex, high-intent AI queries.
Composable and resilient infrastructure (the nervous system): Retailers need to move towards a Headless Commerce model, if they are not already on it. Exposing core logic through domain-driven APIs and MCP servers, ensures your brand remains accessible to any platform, be it ChatGPT, Gemini, or a niche personal stylist agent, without requiring a total architectural overhaul.
The agentic interaction layer: Mastering the handshake
Once the foundation is set, the interaction layer provides the active capabilities for autonomous commerce:
Protocol gateway and orchestration (the handshake): As the universal adaptor for the delegated economy, this block enables secure communication via emerging standards. It ensures only trusted agents can access sensitive pricing or inventory data. It manages the cryptographic handshake that precedes every autonomous sale.
Multimodal brand CX (the voice): Own the customer relationship across three interaction models: agent-to-consumer, business-to-agent and agent-to-agent.
Agentic governance, policy and measurement (the OS): Implement an agentic operating system. Provide the policies, guardrails, observability and ROI tracking necessary to keep autonomous actors safe, legal and brand-aligned.
Accelerating the shift: From engineering foundation to industry outcomes
We recognize that no enterprise builds on a clean slate. To bridge the execution gap, you must take ownership of your digital evolution. By establishing unified agentic control planes and leveraging commerce accelerators embedded with proven best practices, you can speed up the most critical transformation points:
Accelerating discovery: Speed up the enrichment of product catalogs. Normalize identifiers and inject machine-readable metadata so your products are instantly discoverable by AI shoppers.
Accelerating modernization: Decompose legacy monoliths. Reverse-engineer monolithic codebases into the domain-driven, agent-navigable microservices required for a headless future.
Accelerating trust: Adopt secure interaction protocols. Implement the gateways and handshake logic needed to allow agents to research and execute transactions safely.
Accelerating experience models: Deploy the three essential agentic relationships:
A2C (Agent-to-Consumer): Branded concierges that replace static navigation with reasoning.
B2A (Business-to-Agent): Exposing specialized tools and "hooks" so external bots can interact with your brand.
A2A (Agent-to-Agent): Enabling your brand's agents to negotiate directly with a customer's personal agent.
Accelerating multi-agent orchestration: Utilize blueprints for multi-agent orchestration across different clouds and tech stacks. Ensure specialized agents work in concert to solve complex customer goals across a fragmented ecosystem.
Accelerating performance: Optimize agentic reasoning. Provide stress-testing environments to ensure your autonomous experiences are accurate, brand-aligned and free from hallucinations.
Accelerating governance and monitoring: Standardize how you build, govern and operate agents in production. This foundation solves two enterprise blockers simultaneously. It enables the creation of agents that deliver immediate business value while ensuring they deploy safely within rigorous enterprise governance boundaries.
Taking control of your digital future
The transition to agentic commerce is an inevitable structural shift where AI agents are becoming the primary interface for consumers. To avoid exclusion and maintain brand control, retailers must immediately build foundations for visibility, interpretability and transactability to meet customer needs & goals. Thoughtworks facilitates this shift using four core "Product Concepts" designed for the "Delegate and Verify".
Intent-driven discovery: Shifting from static keyword search to high-fidelity, contextual recommendations where your brand is always findable and accurately cited by AI platforms.
Conversational commerce: Replacing legacy navigation with a continuous reasoning layer, where dynamic bundles and pricing are negotiated in real-time based on deep customer context. Customers to be targeted with integrated offers rather than a set of products.
Autonomous transactions: Enabling secure, "Zero-Click" execution where trusted agents finalize payments and logistics within your predefined mandates.
Post-purchase self-healing: Redefining the value chain with agents that proactively manage returns, track shipments and resolve anomalies before the human customer even identifies a problem.
Bridging the gap to foundational readiness is the first step on your journey to an agent-driven future. This ensures your strategy relies on robust, custom engineering.
The question is no longer whether this shift will happen, but how quickly you respond.
