On 8 October 2026, the La Retail Tech community brought together leading retail firms to discuss the challenges facing marketplaces. As a technology and AI partner and an active member of this ecosystem, Novity took part in the discussions.
Leaders from retailers such as But, Conforama, Fnac Darty, Rue du Commerce and Adeo, formerly Groupe Leroy Merlin, pointed to a shared challenge.
Shopping habits are changing. Competition is rising, and customers expect a smooth experience across online and physical channels.
To keep up, IT teams must use AI with care and manage large data volumes. They also need to show a clear return on investment.
Unified commerce and the shift towards hybrid retail models
Online-only retail has its limits, especially in sectors such as home furnishings. Combining digital channels with physical stores is becoming more important.
During the event, Alexandre Falck, the CEO of But and Conforama, said that e-commerce makes up 12% to 15% of their business.
The home furnishings market is closely tied to changes in the property sector. Physical stores therefore remain important. They support logistics and give customers confidence before they buy.
Marketplaces help retailers connect these channels. At Adeo, furniture has become the top online sales category, even though it takes up less space in physical stores. Its marketplace lets the company offer a wider range and support customers throughout the buying journey.
Retailers are also exploring new growth paths. Fnac Darty is expanding its refurbished product offer to capture opportunities in the circular economy, including C2C and B2B2C models. Rue du Commerce is moving further into home products, while Adeo is building a services marketplace that connects customers with tradespeople.
These changes need resilient platforms that prioritise quality and reliability across digital services. At Novity, our engineering teams already support a major French company that specialises in refurbished technology products. That experience gives us direct insight into the technical limits of this market.
As marketplaces grow, their architectures must handlecustomer identity, financial transactions and payments across the entire buying flow. We recently explored these issues with the Purse teams. Their experience shows why flexible transaction systems matter. Without them, new services can quickly create technical and regulatory problems.
Scaling artificial intelligence across retail operations
Generative AI creates new ways to improve customer experience. However, deploying it at scale also raises important governance questions for IT leaders. One concern is the growing use of unapproved AI tools, often called Shadow IT. These tools can expose sensitive business data and raise security risks.
For business leaders, clear rules for AI use are now a priority. Managing the cost of large language models (LLMs) is another challenge.
As discussed during the event, two similar requests can use very different numbers of tokens from one week to the next.
This makes AI infrastructure costs hard to predict and manage, reinforcing the need to rethink the value of traditional IT services in the age of AI.
Three practical AI applications for retail operations
To create lasting value from AI investments, retail leaders are focusing on three areas:
- Supply chain management: Connect supply flows and improve how products are routed to warehouses and stores.
- Software engineering: Speed up development with AI-assisted coding, bug detection and automated test generation.
- Customer service: Introduce chat AI agents that can understand customer intent more accurately than traditional rule-based chatbots.
Moving beyond AI pilots with a technology partner
Scaling AI needs a clear technical framework and a reliable software architecture, supported by a technology and product engineering partner. That is the approach we use when we support clients with data and AI transformation.
Our teams have proved their technical expertise through Anthropic's Claude Certified Architect -- Foundations (CCA-F) certification programme.Novity has also achieved Certified status within the Claude Partner Network.
This recognition shows our focus on structured, well-documented AI engineering. From early planning to production deployment, our specialists review existing architectures and find ways to improve product performance.
We build AI solutions that fit with our clients' existing systems, tools and databases. In practice, this includes autonomous AI agents and solutions built with LLMs, Retrieval-Augmented Generation (RAG) and Model Context Protocol (MCP). The goal is to turn AI capabilities into practical tools that deliver measurable value for e-commerce platforms.
Improving product catalogue data for more accurate AI
Agentic AI is opening new opportunities for chat commerce. However, autonomous assistants can only make useful recommendations when they have access to reliable product data.
Poor-quality data reduces the accuracy of AI recommendations. It can also hurt customer trust and slow adoption. For marketplaces, better product information management (PIM) is therefore a technical priority.
We recently discussed the importance of catalogue optimisation with SteerFox, a specialist in e-commerce customer acquisition and product assortment optimisation. Their analysis shows that businesses need to standardise data flows and automate product information enrichment before introducing AI into customer-facing tools.
Antoine Pougeas, Chief Sales Officer at Novity −
« The quality of the discussions at La Retail Tech Thursdays shows how mature the ecosystem has become. Hearing retail leaders share their experience of managing AI was particularly valuable.
Unpredictable token costs and Shadow IT are real challenges at an organisational level.
This confirms what we see among our clients: technology is moving quickly, but without clear governance and reliable data architecture, costs can spiral.Supporting teams through this transition requires a structured approach.»
Key technology challenges for E-commerce platforms in 2026






