Introduction

In January 2026, the French Competition Authority (FCA) launched a market investigation into conversational agents and agentic commerce[1]. This initiative marks the first time the FCA is examining the downstream segment of the artificial intelligence (AI) value chain, following two earlier studies focused on upstream layers, namely the development of AI models[2] and the energy and environmental impact of AI[3].

As firms increasingly deploy AI applications, competition authorities worldwide are paying closer attention to competitive dynamics in downstream markets. Authorities in Italy[4], Brazil[5], and the European Union have[6], for instance, opened antitrust investigations into Meta concerning its alleged refusal to grant third-party AI assistants access to WhatsApp while integrating its own assistant, Meta AI.

AI assistants that initially provided generated responses are rapidly evolving into conversational agents capable of autonomously performing tasks on users’ behalf, such as shopping. This evolution is contributing to a broader shift from a human-centric web to an agentic web, and from traditional e-commerce to agentic commerce. According to a Morgan Stanley report, agentic commerce could account for between USD 190 billion and USD 385 billion in e-commerce spending in the United States by 2030, representing approximately 10–20% of the market^[7].

This article provides an overview of the FCA’s market study. It first examines the emergence of conversational agents, then analyses the development of agentic commerce, and concludes with research recommendations for the FCA.

Conversational Agents

The FCA’s study focuses on several interrelated issues: the integration of advertising into conversational agents, their embedding within existing digital services, the partnerships formed by agent developers, and the gradual platformisation of conversational agents. It does not study the relationship between conversational agents and search engines. This section outlines recent market developments relevant to these issues.

France is among the countries with the highest adoption of AI tools. According to The Economist, citing data from Microsoft’s AI for Good Lab, more than 40% of the French working-age population used conversational agents in 2025[8]. Médiamétrie data indicate that, as of September 2025, the leading services in France were OpenAI’s ChatGPT (21.6 million unique visitors), followed by Google Gemini (2.8 million), Mistral AI’s Le Chat (1.5 million), Perplexity (1.3 million), and Microsoft Copilot (1 million). French users rely on conversational agents for a wide range of educational, professional, and personal uses[9].

Competition in this sector is fierce and largely driven by rapid innovation. Using ChatGPT as an illustration, recent developments include domain-specific functionality (such as education, health, or shopping), expanded content-generation capabilities (including images and videos), search-related queries (including web search and deep research), and greater personalisation (such as keeping conversational context). Personalisation is particularly relevant from a competition perspective, as developers may collect and process user-generated data to improve performance, subject to user consent. As noted by Professors Hagiu and Wright, such improvements operate at both the model level, benefiting all users, and the user level, delivering personalised outputs. This dual mechanism may generate positive feedback loops, in which greater use leads to improved performance and further user adoption[10]. However, empirical evidence on the competitive impact of such data feedback loops is still lacking.

Distribution strategies also play a central role in competitive dynamics. Large online platforms often integrate their own conversational agents into existing services, such as Meta AI within Instagram, Facebook, and WhatsApp, or Google Gemini as both a standalone and integrated service within Google Search. Other developers, such as OpenAI and Mistral, primarily rely on standalone distribution. They also distribute through partnerships with device manufacturers (e.g., Google/Samsung[11]) and telecommunications operators (e.g., Mistral/Free[12]). Notably, OpenAI’s ChatGPT remains the most widely used service in France, suggesting that vertical integration does not necessarily guarantee a competitive advantage, but that first-mover does[13].

Finally, conversational agents are increasingly becoming platforms. To enhance user value, developers are enabling business users to reach end users directly through conversational interfaces. For example, ChatGPT hosts an application store that allows users to interact with services such as OpenTable[14], and it enables partner retailers, such as Instacart in the United States of America (USA), to facilitate direct purchases with the chat in exchange for a commission[15]. Google offers similar functionality through Gemini and AI Mode to partner retailers, such as Shopify and Walmart in the USA[16]. This platformisation may give rise to indirect network effects, although their magnitude remains empirically underexplored.

These developments have implications for intermediaries and online advertising. The Australian Competition and Consumer Commission (ACCC) has observed that AI agents may alter consumer interactions with intermediary platforms, including online marketplaces, by enabling more direct engagement with business users[17]. Professor Manganelli suggests that conversational agents could evolve into “meta-gatekeepers,” potentially disintermediating existing platforms and concentrating decision-making power[18]. At the same time, incumbents are adapting by integrating agentic features into their own services, such as Google’s Business Agents in the USA, which enable end users to transact directly with an agent tailored to the business user’s voice, from the queries to checkout[19]. In advertising markets, traditional formats, such as advertising banners and sponsored results, designed for human attention may lose effectiveness if AI agents disregard these contents. While evidence of such impact remains limited[20], advertising models are adapting. For instance, Google offers advertising within the generated answer[21]. It is also experimenting with direct offers in the USA, which allows advertisers to present exclusive offers to end users, such as discounts[22].

Agentic Commerce

Given its potential economic impact, the FCA devotes particular attention to agentic commerce. The authority examines the structure of the value chain, the relationships between its constituent services, its interaction with traditional e-commerce, and the associated competitive challenges. This section provides an overview of the technical and commercial relationships surrounding agentic commerce.

From a technical perspective, agentic commerce requires AI agents to interact autonomously with merchants, content providers, payment systems, and other agents. This has spurred the development of new services and protocols, including the Universal Commerce Protocol (UCP) and the Agent Payments Protocol (AP2). Developed by Google in collaboration with industry partners such as Shopify, Etsy, Wayfair, Target, and Walmart, the UCP is an open-source standard designed to enable seamless commerce journeys across consumer interfaces and payment providers^[23]. AP2 facilitates secure agent-based payments across platforms^[24]. Other open-source initiatives, such as Agent2Agent (A2A)^[25] and the Model Context Protocol (MCP)^[26], respectively support interoperability between agents and external systems. Together, these developments signal a transition toward an agentic web^[27]. From a competition perspective, standard-setting processes are critical, as standards may influence market access, innovation, and competitive conditions, requiring vigilance that they do not lead to anticompetitive practices[28]. 

Commercially, agentic commerce is reshaping the e-commerce landscape. AI agent developers such as OpenAI, Google, Microsoft, and Perplexity increasingly compete with online marketplaces, including Amazon, while simultaneously stimulating innovation. E-commerce platforms, such as Shopify, and marketplaces, such as Amazon, are also expanding their reach through agentic features. Shopify’s Agentic plan allows non-Shopify merchants to sell products through AI channels, increasing brand visibility across multiple sales environments[29]. Similarly, Amazon’s “Buy for Me” feature enables users in the USA to purchase products from non-Amazon merchants through the Amazon Shopping app[30]. While such practices may enhance consumer choice, brand visibility, and sales, they have also raised concerns among some non-Amazon merchants about unauthorised scraping, legal exposure, and reputational risks^[31].

Recommendations

The FCA’s market study takes place in a context of rapid technological, market, and regulatory developments that are likely to shape both the deployment of AI agents and the evolution of competition in digital markets. At this preliminary stage, two areas appear particularly relevant for the authority’s examination.

Recommendation 1: Examine the Impact of Digital Regulations in AI Markets

Digital regulations, notably the Digital Markets Act (Regulation (EU) 2022/1925, DMA), shape the conduct of designated gatekeepers in several digital services, potentially impacting their development and deployment of AI agents.

The FCA should consider whether, and to what extent, existing obligations may affect the distribution, integration, and monetisation of AI agents, as well as the competitive conditions faced by non-gatekeeper developers. Such an assessment would help clarify whether current regulatory tools are neutral, supportive, or potentially constraining with respect to innovation and competition in conversational agents and agentic commerce.

Recommendation 2: Assess the Implications of AI Agents for the Analytical Framework of Digital Markets

AI agents may alter competitive dynamics in several digital markets traditionally analysed under competition law, including e-commerce, online intermediation, search, web browsers, and online advertising. In this context, the FCA should explore whether conversational agents and agentic commerce alter analytical frameworks, notably with respect to market definition, entry and expansion conditions, and potential theories of harm

[1] Autorité de la concurrence, Conversational Agents: The Autorité Starts Inquiries Ex Officio with a View to Issuing an Opinion, 9 January 2026 (accessed 15 January 2026). Available at: https://www.autoritedelaconcurrence.fr/en/article/conversational-agents-autorite-starts-inquiries-ex-officio-view-issuing-opinion

[2] Autorité de la concurrence, Opinion 24-A-05 of June 28, 2024, On the Competitive Functioning of the Generative Artificial Intelligence Sector, 28 June 2024 (accessed 15 June 2026). Available at: https://www.autoritedelaconcurrence.fr/en/opinion/competitive-functioning-generative-artificial-intelligence-sector

[3] Autorité de la concurrence, Competition Issues Surrounding the Energy and Environmental Impact of Artificial Intelligence (AI), 17 December 2025 (accessed 15 January 2026). Available at: https://www.autoritedelaconcurrence.fr/sites/default/files/2026-01/Study_AI_Energy_ADLC.pdf

[4] Autorità Garante della Concorrenza e del Mercato A576 - Meta AI: The Italian Competition Authority Orders Meta to Suspend the Terms Excluding Competing AI Chatbots from WhatsApp, 24 December 2025 (accessed 15 January 2026). Available at: https://en.agcm.it/en/media/press-releases/2025/12/A576

[5] Conselho Administrativo de Defesa Econômica, Cade Abre Inquérito Contra Meta e Aplica Medida Preventiva Suspendendo Novos Termos do Whatsapp Sobre IA, 12 January 2026 (accessed 15 January 2026) (only available in Portuguese). Available at: https://www.gov.br/cade/pt-br/assuntos/noticias/cade-abre-inquerito-contra-meta-e-aplica-medida-preventiva-suspendendo-novos-termos-do-whatsapp-sobre-ia

[6] Commission Opens Antitrust Investigation into Meta's New Policy Regarding AI Providers' Access to WhatsApp, European Commission, 4 December 2025 (accessed 16 January 2026). Available at: https://ec.europa.eu/commission/presscorner/detail/en/ip_25_2896

[7] Here Come the Shopping Bots, Morgan Stanley, 8 December 2025 (accessed 14 January 2026). Available at: https://www.morganstanley.com/insights/articles/agentic-commerce-market-impact-outlook

[8] Which Countries are Adopting AI the Fastest?, The Economist, 12 January 2026 (accessed 15 January 2026). Available at: https://www.economist.com/graphic-detail/2026/01/12/which-countries-are-adopting-ai-the-fastest

[9] Young People at the Forefront of the Conversational AI Revolution, Mediametrie, 30 October 2025 (accessed 15 January 2026). Available at: https://www.mediametrie.fr/en/young-people-forefront-conversational-ai-revolution

[10] Andrei Hagiu and Julian Wright, Artificial Intelligence and Competition Policy, International Journal of Industrial Organization, 2025 (accessed 16 January 2026). Available at: https://www.sciencedirect.com/science/article/pii/S0167718725000013

[11] Samsung and Google Cloud Join Forces to Bring Generative AI to Samsung Galaxy S24 Series, Samsung Blog, 18 January 2026 (accessed 15 January 2026). Available at: https://www.samsungmobilepress.com/articles/samsung-and-google-cloud-join-forces-to-bring-generative-ai-to-samsung-galaxy-s24-series

[12] Free France's First Telco to Include a Premium AI Assistant in its Mobile Plans Le Chat Pro Developed by Mistral AI, Free Blog, 10 February 2025 (accessed 15 January 2026). Available at: https://www.iliad.fr/en/actualites/article/free-france-s-first-telco-to-include-a-premium-ai-assistant-in-its-mobile-plans-le-chat-pro-developed-by-mistral-ai

[13] Though, in France, Google Gemini is not yet integrated within Google Search via Google AI Overviews and Google AI Mode, as these services are not yet available to French users.

[14] Introducing the GPT Store, OpenAI, 10 January 2024 (accessed 16 January 2026). Available at: https://openai.com/index/introducing-the-gpt-store/

[15] Instacart and OpenAI Partner on AI Shopping Experiences, OpenAI, 8 December 2025 (accessed 16 January 2026). Available at: https://openai.com/index/instacart-partnership/

See also, Buy It In ChatGPT: Instant Checkout and the Agentic Commerce Protocol, OpenAI, 29 September 2025 (accessed 16 January 2026). Available at: https://openai.com/index/buy-it-in-chatgpt/

[16] Vidhya Srinivasan, New Tech and Tools for Retailers to Succeed in an Agentic Shopping Era, Google Blog, 11 January 2026 (accessed 16 January 2026). Available at: https://blog.google/products/ads-commerce/agentic-commerce-ai-tools-protocol-retailers-platforms/

[17] Australian Competition and Consumer Commission, Recent Developments in Artificial Intelligence Industry Snapshot, 17 December 2025 (accessed 16 January 2026). Available at: https://www.accc.gov.au/about-us/publications/recent-developments-in-ai-industry-snapshot

[18] Antonio Manganelli, Foundation Models and Generative AI Applications: What Competitive Concerns?, Forthcoming in European Competition Journal, 2025 (accessed 16 January 2026). Available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5242028

[19] Vidhya Srinivasan, New Tech and Tools for Retailers to Succeed in an Agentic Shopping Era, Google Blog, 11 January 2026 (accessed 16 January 2026).

[20] For a study in the e-commerce sector, see, Amine Allouah et al., What Is Your AI Agent Buying? Evaluation, Implications and Emerging Questions for Agentic E-Commerce, 2025 (accessed 16 January 2026). Available at: https://arxiv.org/abs/2508.02630

[21] More Opportunities for your Business On Google Search, Google Blog, 21 May 2025 (accessed 16 January 2026). Available at: https://blog.google/products/ads-commerce/google-search-ai-brand-discovery/

[22] Vidhya Srinivasan, New Tech and Tools for Retailers to Succeed in an Agentic Shopping Era, Google Blog, 11 January 2026 (accessed 16 January 2026).

[23] Amit Handa and Ashish Gupta, Under the Hood: Universal Commerce Protocol (UCP), Google Blog, 11 January 2026 (accessed 14 January 2026). Available at: http://developers.googleblog.com/under-the-hood-universal-commerce-protocol-ucp/

[24] Stavan Parikh and Rao Surapaneni, Powering AI Commerce with the New Agent Payments Protocol (AP2), Google Blog, 16 September 2025 (accessed 14 January 2026). Available at: https://cloud.google.com/blog/products/ai-machine-learning/announcing-agents-to-payments-ap2-protocol?hl=en

[25] Agent2Agent (A2A) Protocol (accessed 14 January 2026). Available at: https://a2a-protocol.org/latest/

[26] What is the Model Context Protocol (MCP)? (accessed 14 January 2026). Available at: https://modelcontextprotocol.io/docs/getting-started/intro

[27] Yingxuan Yang et al., Agentic Web: Weaving the Next Web with AI Agents, 2025 (accessed 14 January 2026). Available at: https://arxiv.org/abs/2507.21206

[28] OECD, Standard Setting, Series Roundtables on Competition Policy, 2010 (accessed 16 January 2026). Available at: https://www.oecd.org/content/dam/oecd/en/publications/reports/2011/03/standard-setting_24c2823d/8b423811-en.pdf

[29] The Agentic Commerce Platform: Shopify Connects any Merchant to Every AI Conversation, Shopify, 11 January 2026 (accessed 16 January 2026). Available at: https://www.shopify.com/news/ai-commerce-at-scale

[30] Amazon’s New 'Buy for Me' Feature Helps Customers Find and Buy Products from Other Brands’ Sites, Amazon Blog, 3 April 2025 (accessed 16 January 2026). Available at: https://www.aboutamazon.com/news/retail/amazon-shopping-app-buy-for-me-brands

[31] Allison Smith, Brands Are Upset That ‘Buy for Me’ is Featuring Their Products on Amazon Without Permission, Modern Retail, 6 January 2026 (accessed 15 January 2026). Available at: https://www.modernretail.co/technology/brands-are-upset-that-buy-for-me-is-featuring-their-products-on-amazon-without-permission/