Agentic AI in Retail: True Agency Requires Customer Sovereignty

The Current State: Efficiency Disguised as Agency. When will the value switch to the consumer, rather than the seller? A mono-post this week on whether the customer really is queen.

The retail industry is witnessing an unprecedented wave of AI announcements, with Google's "Buy with Google," Amazon's withdrawal from Google Shopping, and bold predictions about the "death of the web." Yet amid these panting releases and amped-up pronouncements, most of the value to the non-ai-afficionado is merely sophisticated efficiency, not true agency.

Sovereign AI © iJindal’s chats with Perplexity ;)

Google's recent AI Mode with "Buy for Me" functionality exemplifies this disassociation between the agentic claims and reality. When Google tracks a product's price and automatically purchases it using your stored payment details, it's undoubtedly convenient. But who truly benefits from this streamlined process? The answer reveals the fundamental flaw in current implementations: the agent primarily serves the platform's interests, not the customer's. Furthermore, Google is the platform that has made shopping granular to the point that a searcher (oops, human customer) enters a click-frenzy of open tabs, mental overhead, vendor comparison, delivery juggling and energy-sapping keyboard-tapping. No doubt, streamlining the process is a civil service, but the absence of pain is not the same as joy.

The absence of pain is not the same as joy

Amazon's dramatic exit from Google Shopping ads in July 2025 has had pundits and experts mulling over the reason. It’s beyond my understanding or knowledge as to “why” (although Digiday has a good roundup) but for the customer the reason doesn’t matter. From that perspective it lays bare the battle between the giants to influence (and therefore take a cut of) your spending. Google has been the grand conductor of intent, funnelling and pointing every question towards a monetisable click. Amazon has been the universal “get it now” engine.

Every product you can name is available under your thumb from Amazon. Where you don’t know the exact item, Google is there. Where you didn’t know you even desired a thing, thank you Instagram and TikTok.

So, that’s a long way of saying that these announcements to grease the process between opening your eyeball and spending are all very laudable, but they don’t mention the customer anywhere.

Beyond Convenience: The Case for Customer-Centric Agency

Consumers will only see value in AI when the AI-powered systems work wholly for the customer, not as intermediaries serving platform interests. This requires a fundamental shift from platform-controlled data and transactions to customer-sovereign digital identities.

I have covered Sovereign AI in a note here:

Customer-centred worldview, not vendor-based

Current agentic systems suffer from fragmented visibility. Amazon's agent sees your Amazon data, Google's sees your Google activity, but neither possesses a complete view of your preferences, financial situation, or life circumstances. This fragmentation leads to suboptimal recommendations and decisions that may benefit the platform's bottom line more than the customer's welfare.

Real agentic AI requires access to your complete digital profile: your financial health, spending patterns across all platforms, personal values (sustainability preferences, ethical considerations), life stage (new parent, retiree, student), and even health considerations that might influence purchasing decisions. Only with this comprehensive view can an AI agent make truly optimal decisions on your behalf.

But whom do you trust with this view?

For customers to grant such comprehensive access, they need ironclad assurance that their agent works solely in their interest. This creates a fundamental trust infrastructure problem that current platform-based solutions cannot credibly solve.

Research shows that 79% of consumers would stop interacting with brands they don't trust to safeguard personal data, while 74% believe their personal data is more vulnerable than ever¹. The rise of agentic AI, requiring unprecedented data access, has caused industry insiders to share concerns about privacy, but the reality is that governments are offering health data to private US firms and consumers haven’t yet railed against AI. It’s going to take some major leaks and failures before the stable door is retrospectively closed to stop the spilled milk escaping.

Two Pathways to Customer-Sovereign Agency

Customers clearly cannot invent a global framework on their own, so the question really is “which encompassing system(s) will become the norm(s)”. Governments are well-placed to undertake this - they are of and for the people, after all! There may also be trusted companies that already know lots about our most intimate facts - Apple (has my life in its ecosystem), the National Health Service, perhaps a legacy bank…

National Digital Identity Frameworks

Several countries are pioneering comprehensive digital identity systems that could serve as the foundation for truly customer-centric agentic commerce:

Estonia's Success Story: With over two decades of experience, Estonia's national e-ID system demonstrates the viability of government-backed digital identity. The system enables citizens to vote online, sign documents digitally, and access services seamlessly—all while maintaining control over their data. (See Regula Forensics for a handy list of national digital IDs).

India's Scale: Aadhaar (the Unique Identification Authority of India) has enrolled over 1.29 billion residents, with 95% of adults using it monthly and 90% reporting satisfaction. This massive adoption shows that citizens will embrace digital identity systems when they perceive genuine benefit.

Europe's Ambitious Vision: The EU's eIDAS 2.0 regulation mandates that all member states provide digital identity wallets by November 2026, serving 450 million citizens. Unlike current platform solutions, these wallets will be "controlled by users" with the principle that "anything which is not necessary to share will not be shared"².

The European approach is particularly compelling because it explicitly prioritises user control. Citizens will "choose and keep track of their identity, data and certificates which they share with third parties". This user-centric approach creates the foundation for truly customer-sovereign agentic commerce.

That said, ‘user control’ is both a challenge to implement and manage and in some quarters is seen as code to allow protectionist measures against the US-based global tech companies. At this early stage we can only comment on the proposal and it’s very much at the forefront of the consumer-centred worldview.

Apple's Privacy-First Ecosystem

Another viable route is that an existing global firm, that has a large user base, an existing ecosystem, and is not involved in the final-click monetisation world, might gain a de facto role. Apple, having built substantial consumer trust through privacy-first policies, and which doesn’t monetise user data for commerce conversion, might be a candidate.

Apple's recent introduction of Digital ID support through the W3C Digital Credentials API signals a potential move toward becoming a neutral identity provider. Mattr’s article gives details on the move along with analysis of the systems and context. Apple is seen as being more on the side of the Apple customer than any individual shopping access point or gateway. Forrester research and an analysis piece by Martin Gill in March 2025 show that more UK users trust Apple than trust the UK Government (35% to 25%).

Would you trust Apple enough for it to be the underpinning of your sovereign digital identity? If not, is there another non-governmental body you’d trust?

The Data Sovereignty Imperative

Beyond Platform Silos

Current agentic implementations suffer from the "fragmentation challenge". Each platform sees only a slice of customer behaviour, leading to suboptimal decisions (both from a marketer’s perspective and the customer’s).

Customer data ownership must encompass three core principles :

• Protection: Robust security and privacy safeguards

• Control: Customer authority over data usage and sharing

• Monetisation: Ensuring customers, not just platforms, benefit from the value of their data.

This last part is likely to be interesting since at present the main monetisation of data is “give me your data and I’ll spam you with some vouchers or discount codes”. Where a market for realising (and sharing in) data value exists then consumer attitudes will change.

Privacy by Design vs. Surveillance Capitalism

The current trajectory toward agentic commerce raises significant privacy concerns. Agentic AI systems require a great deal of personal data to function effectively, creating unprecedented surveillance demands. Without proper safeguards, these systems risk becoming tools for "surveillance and profiling" rather than customer empowerment. As retailers during the working day we may not care - after all, targets be targets! - but in our private lives the knowledge we have of data scraping, exchange and use may be of concern when we are on the other side of the checkout page, as customers!

The European model offers a compelling alternative through privacy-by-design principles. Digital identity wallets built on eIDAS 2.0 will enable selective disclosure—sharing only the minimum necessary information for each transaction. This approach maintains customer privacy while enabling sophisticated agentic capabilities. However, coming after the EU’s legislation on data privacy (GDPR) and other digital regulations there are concerns that it may slow the pace of development in Europe and draw challenges from the US-based tech companies and indeed the US Government. It’s not yet clear whether the AI hare or the consumer tortoise will “win”, or even whether this is a race at all, but it shows the two different worldviews - the tech companies and marketers versus the consumers’ interests.

Implementation Challenges and Market Reality

The platform-specific agentic systems are ready to use, have an existing and familiar user base and will be heavily promoted. At the same time the technology and framework for Sovereign AI is also feasible and so the two determinants of adoption are likely to be promotion/awareness and ‘trust’. A weariness with platform lock-in, growing impact of regulatory changes, and increased familiarity with national digital identities could give some momentum to a consumer shift - especially if this is made easy.

What True Agency Looks Like

Imagine an agentic AI system that:

• Accesses your complete financial picture across all institutions to recommend purchases that fit your budget and financial goals

• Considers your health data when suggesting food products or supplements

• Factors in your sustainability preferences and ethical values

• Negotiates prices across all available vendors, not just those in a single ecosystem

• Optimises for your long-term welfare, not platform engagement metrics

• Provides a simple, transparent interface to your choices and consults you where appropriate so it’s not a ‘black box’ but rather a trusted helper.

The Competitive Advantage of Customer Sovereignty

Organisations that embrace customer-sovereign agentic commerce may gain competitive advantages:

• Higher customer trust leading to broader data sharing permissions

• Better decision quality from comprehensive customer understanding

• Reduced platform dependency and associated costs

• Regulatory compliance with evolving privacy requirements

This isn’t an ‘either/or’ decision at this point, but rather a planning approach that keeps Sovereign AI options open, while emphasising that acting with integrity, transparency and authenticity is both an ethical brand decision and a higher-ROI approach in the medium term and longer.

Conclusion: The Choice Ahead

Current implementations of agentic AI offer impressive efficiency gains but fall short of true agency because they primarily serve platform interests rather than customer welfare.

True agentic commerce requires customer data sovereignty - comprehensive digital profiles that customers control and agents that work solely in customer interests. This vision can only be realised through trusted identity frameworks: either national digital identity systems built on privacy-by-design principles or ecosystems like Apple's that have earned customer trust through aligned incentives.

The companies and countries that get this right will define the next era of commerce. Those that cling to platform-centric models, no matter how sophisticated their AI, will deliver mere efficiency while true agency remains elusive.

The question isn't whether agentic AI will transform retail—it already has. The question is whether this transformation will empower customers or simply make their capture more efficient. The answer depends on who controls the data and whose interests the agents ultimately serve.

Your thoughts?

I admit I’m in Apple’s ecosystem, but what other trust players would you suggest? Do you see Sovereign AI as a pipedream of an aging Marxist, or the necessary underpinning of a neo-liberal economy? Let me know your views and how (if at all) you’re planning for the ‘sovereign consumer’.

1. https://www.c-suite-strategy.com/blog/building-digital-trust-key-strategies-for-todays-businesses

2. Useful article on identities here: https://www.retailbankerinternational.com/comment/eu-digital-identity-wallets-steps-for-successful-implementation-2/