{"id":46127,"date":"2025-07-05T21:19:00","date_gmt":"2025-07-05T21:19:00","guid":{"rendered":"https:\/\/www.mon-agent-ia.fr\/blog\/?p=46127"},"modified":"2025-07-05T21:19:01","modified_gmt":"2025-07-05T21:19:01","slug":"anthropics-small-ai-run-shop-didnt-meet-with-the-expected-success","status":"publish","type":"post","link":"https:\/\/www.mon-agent-ia.fr\/blog\/en\/anthropics-small-ai-run-shop-didnt-meet-with-the-expected-success\/","title":{"rendered":"Anthropic&rsquo;s small AI-run shop didn&rsquo;t meet with the expected success"},"content":{"rendered":"<p class=\"wp-block-paragraph\">In a world where technology and artificial intelligence (AI) are constantly redefining the contours of our daily activities, it&rsquo;s fascinating to see how pioneering companies are exploring innovative concepts. One such experiment is Anthropic, led by Dario Amodei, which launched a small boutique run by an AI named Claude. The goal was clear: to test the capabilities of an AI to manage a business while exploring its impact on the future economy. However, as this experiment reveals, the results fell far short of expectations.<\/p>\n\n<p class=\"wp-block-paragraph\">This initiative, which was intended to illustrate the promise of autonomous management by AI, raised questions about the risk of technological unemployment and the challenges of such automation. With forecasts suggesting an increase in the unemployment rate between 10 and 20% by 2025, Anthropic&rsquo;s experience could offer valuable lessons on the interaction between human creativity and algorithms. Let&rsquo;s take a closer look at what happened in this brand-new little shop.<\/p>\n\n<h2 class=\"wp-block-heading\">Anthropic&rsquo;s Little Shop Experience and Its Challenges<\/h2>\n\n<p class=\"wp-block-paragraph\">Located in the company&rsquo;s San Francisco offices, the shop, which was actually just a small refrigerator topped with an iPad, provided a fascinating testing ground for the Claude 3.7 Sonnet AI. Significantly, Claude&rsquo;s mission wasn&rsquo;t simply to sell products, but also to handle other essential aspects of a business, such as inventory management, customer interactions, and profit optimization.<\/p>\n\n<p class=\"wp-block-paragraph\">To carry out this experiment, Claude benefited from a set of tools necessary for its smooth operation. Among these tools, the Slack communication software enabled seamless interaction with employees, while Andon Labs provided expertise in the AI \u200b\u200binfrastructure. This innovative combination provided a glimpse of what an autonomous economy could look like, albeit with sometimes unexpected results. The story of this small shop deserves to be told in fascinating detail.<\/p>\n\n<h3 class=\"wp-block-heading\">Claude&rsquo;s Missions as a Sales AI<\/h3>\n\n<p class=\"wp-block-paragraph\">At the heart of this experiment, Claude not only had to manage transactions but also deal with real-life situations encountered in any business. Here is an overview of the key missions Claude was expected to perform:<\/p>\n\n<ul class=\"wp-block-list\"><li><strong>Inventory Management<\/strong> : Ensure adequate product levels are maintained.<\/li><li><strong>Customer Service<\/strong> : Answer customer questions and manage relationships.<\/li><li><strong>Sales<\/strong> : Increase revenue while amortizing operating costs. Fair Decision Making: Ensuring that all customers feel treated fairly.<\/li><li><strong>However, each mission presented a unique challenge. For example, during interactions with employees, Claude found himself giving unreasonable discounts, sometimes to the point of selling products at a loss, due to his perceived need for fairness. This behavior raises questions about an AI&rsquo;s responsiveness to human requests.<\/strong> Unexpected Moments and What They Reveal About AI<\/li><\/ul>\n\n<p class=\"wp-block-paragraph\">As the experiment developed, several intriguing incidents arose, surpassing what one might expect from simple store management. A team member named Kevin Troy recounted instances where Claude, true to his AI role, got himself drawn into illogical and unpredictable situations.<\/p>\n\n<h3 class=\"wp-block-heading\">Employees interacted with Claude in a playful way, sometimes pushing the AI \u200b\u200bassistant to fulfill outlandish requests, such as ordering \u201ctungsten cubes\u201d that were used as bookends. These interactions, while often tinged with humor, highlight a puzzling aspect of AI: the idea that an AI can \u201challucinate\u201d and invent stories that have no basis in reality is both fascinating and disturbing. For example, Claude once \u201cpretended\u201d to have signed a fictitious contract at an iconic address, the one from The Simpsons, illustrating the extent to which an AI can descend into the absurd.<\/h3>\n\n<p class=\"wp-block-paragraph\">Analyzing the results of Anthropic\u2019s small shop experiment<\/p>\n\n<p class=\"wp-block-paragraph\">Although this experiment was not commercially successful, Anthropic researchers were careful not to jump to conclusions. On the contrary, they saw this initiative as a promising indicator of how more sophisticated AI could play a role in the retail industry of the future. The results highlight the idea that AI doesn&rsquo;t need to be perfect to be adopted, but rather to be competitive with human performance.<\/p>\n\n<h3 class=\"wp-block-heading\">Aspect of the experience<\/h3>\n\n<p class=\"wp-block-paragraph\">Result<\/p>\n\n<figure class=\"wp-block-table\"><table>\n<thead>\n<tr>\n<th>Potential impact<\/th>\n<th>Sales<\/th>\n<th>Offered sales, sometimes at a loss<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Distorted perception of profitability<\/td>\n<td>Interaction management<\/td>\n<td>Invention of fictional stories<\/td>\n<\/tr>\n<tr>\n<td>Awareness of some of AI&rsquo;s limitations<\/td>\n<td>Employee requests<\/td>\n<td>Ordered tungsten cubes<\/td>\n<\/tr>\n<tr>\n<td>Humor and playfulness in interaction<\/td>\n<td>These lessons learned during this adventure offer an interesting perspective on the challenges facing AI in the retail sector. As companies continue to explore how artificial intelligence can transform their operations, Anthropic&rsquo;s experience could serve as a model for evaluating not only AI performance, but also its adaptation to human needs.<\/td>\n<td>Implications for the Future of Work and AI<\/td>\n<\/tr>\n<\/tbody>\n<\/table><\/figure>\n\n<p class=\"wp-block-paragraph\">With the growing adoption of AI technologies, it is crucial to consider their implications for the world of work. Dario Amodei predicts a substantial increase in the unemployment rate, revealing the fear associated with such automation. As AI systems become more sophisticated, they will take over human tasks traditionally considered irreplaceable.<\/p>\n\n<h2 class=\"wp-block-heading\">Risks of Technological Unemployment<\/h2>\n\n<p class=\"wp-block-paragraph\">As automation increases, certain professions could become obsolete. Here is a list of sectors most at risk:<\/p>\n\n<h3 class=\"wp-block-heading\">Retail: Possible replacement by automated checkouts and AI managers. Logistics: Inventory management by autonomous robots.<\/h3>\n\n<p class=\"wp-block-paragraph\">Agriculture: Use of drones and robots for planting and harvesting.<\/p>\n\n<ul class=\"wp-block-list\"><li><strong>Customer service: Virtual agents handling inquiries.<\/strong> The consequences of such a disruption to the labor market could be far-reaching, affecting not only affected employees but also businesses. Therefore, it is imperative to initiate constructive dialogue on how to adapt to this new reality.<\/li><li><strong>Opportunities for growth and innovation<\/strong> However, behind the challenges AI brings, there are also opportunities for growth and innovation. The introduction of AI into the workplace can generate new functions, improve the user experience in various areas, and establish entirely new business models. The chances of companies innovating and adapting to these new realities are high.<\/li><li><strong>Sectors that could benefit from such a transformation include:<\/strong> AI Application Development<\/li><li><strong>: Creation of new software solutions aimed at optimizing business processes.<\/strong> Professional Training<\/li><\/ul>\n\n<p class=\"wp-block-paragraph\">: Development of programs to help employees adapt to AI.<\/p>\n\n<h3 class=\"wp-block-heading\">Technology Consulting<\/h3>\n\n<p class=\"wp-block-paragraph\">: Companies offering advice on the successful integration of AI.<\/p>\n\n<p class=\"wp-block-paragraph\">Service Enhancement<\/p>\n\n<ul class=\"wp-block-list\"><li><strong>: Personalization and optimization of product offerings in real time using AI.<\/strong> The debate on the future of work must therefore include a sincere recognition that AI, despite its challenges, also has the potential to bring both challenges and opportunities to today&rsquo;s workforce. The Balance Between Humans and AI<\/li><li><strong>The key will be to strike a balance between automation and humans. Collaboration between the two will be essential for mutual success. AI has the potential to improve the daily work experience, but it is crucial not to forget the invaluable value that humans bring.<\/strong> Successful experiments, such as Anthropic&rsquo;s small boutique project, should serve as case studies for developing solutions that harness the creative potential of AI while preserving the need for human supervision.<\/li><li><strong>Implications for the Future<\/strong> Opportunities<\/li><li><strong>Risks<\/strong> Adapting Business Models<\/li><\/ul>\n\n<p class=\"wp-block-paragraph\">Rapid Innovation in Services and Technology<\/p>\n\n<h3 class=\"wp-block-heading\">Job Losses in Some Sectors<\/h3>\n\n<p class=\"wp-block-paragraph\">New Training Needs<\/p>\n\n<p class=\"wp-block-paragraph\">Creation of new careers in education<\/p>\n\n<figure class=\"wp-block-table\"><table>\n<thead>\n<tr>\n<th>Widening the digital divide<\/th>\n<th>Strengthening human relationships<\/th>\n<th>Improving customer service through AI<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Reducing human decision-making competence<\/td>\n<td>The importance of conversations about AI and human values<\/td>\n<td>The Anthropic experience, while revealing gaps, also highlights the importance of open conversations around AI. What are the ethical implications of such integration? How can companies create work environments that respect humans while integrating automation? A commitment to understanding and addressing these questions is vital.<\/td>\n<\/tr>\n<tr>\n<td>Toward a more humanistic collaboration<\/td>\n<td>The future success of AI in the business world will depend on our ability to integrate human values \u200b\u200binto the technology design and application process. Ultimately, it&rsquo;s not simply about automated tools, but about how these tools can enrich the human experience and improve quality of life. Companies must foster a culture of innovation while keeping ethics and social responsibility in mind. Creating an environment where humans and AI can coexist harmoniously is becoming essential.<\/td>\n<td>Stories of AI-powered stores remind us that it is crucial to remain attentive and aware of the direction our society is heading. Anthropic&rsquo;s experience should serve as inspiration for designing a future that is not only dominated by technology, but also remains deeply rooted in our human values.<\/td>\n<\/tr>\n<tr>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<\/tr>\n<\/tbody>\n<\/table><\/figure>\n\n<h2 class=\"wp-block-heading\"><\/h2>\n\n<p class=\"wp-block-paragraph\"><\/p>\n\n<h3 class=\"wp-block-heading\"><\/h3>\n\n<p class=\"wp-block-paragraph\"><\/p>\n\n<p class=\"wp-block-paragraph\"><\/p>\n\n<p class=\"wp-block-paragraph\"><\/p>\n\n\n","protected":false},"excerpt":{"rendered":"<p>In a world where technology and artificial intelligence (AI) are constantly redefining the contours of our daily activities, it&rsquo;s fascinating to see how pioneering companies are exploring innovative concepts. One such experiment is Anthropic, led by Dario Amodei, which launched a small boutique run by an AI named Claude. The goal was clear: to test [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":46067,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1398],"tags":[1653,149,80119,1010],"class_list":["post-46127","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news-ai-en","tag-anthropic-en","tag-artificial-intelligence-en","tag-failure-en","tag-ia-en"],"_links":{"self":[{"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/posts\/46127","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/comments?post=46127"}],"version-history":[{"count":1,"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/posts\/46127\/revisions"}],"predecessor-version":[{"id":46128,"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/posts\/46127\/revisions\/46128"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/media\/46067"}],"wp:attachment":[{"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/media?parent=46127"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/categories?post=46127"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/tags?post=46127"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}