AI for B2B Commerce: adoption and evolution

While many companies still see AI mainly as a tool for automating repetitive tasks, the latest evolution points toward agentic shopping: AI systems capable not only of assisting with research, but also of autonomously completing commercial transactions based on user-defined criteria. 

This approach is already finding concrete applications in both B2C and B2B contexts. Many B2B companies are implementing open e-commerce platforms accessible without mandatory login, where products are displayed with detailed descriptions even when prices are not publicly visible. This openness makes them ideal for initiating conversations with AI systems, supporting use cases ranging from automated lead generation to order automation. 

To fully leverage this potential, however, companies must work on content quality: improving product descriptions, enriching technical information and structuring data so that AI engines can correctly interpret and surface it within relevant search contexts. 

At this point, one might ask: what is the difference between an advanced CRM and AI-enabled Agentic Commerce? The answer lies in how data is processed and used. 

Traditional CRM systems work with structured data: names, e-mail addresses, transaction histories, and commercial activities. Companies gather information through various touchpoints and consolidate it into a database that essentially becomes a massive spreadsheet used to find contacts, segment lists or track opportunities. 

AI introduces instead the concept of a vector profile: a multidimensional numerical analysis representing each customer not only through transactional and demographic data, but also through behaviours, preferences, interactions and patterns. This approach allows AI to be queried naturally in order to generate predictive insights, identify similar profiles, automatically enrich data with inferred information and orchestrate actions across multiple systems. 

For example, if a user with specific socio-demographic characteristics purchases a certain type of product, AI can automatically identify other prospects with similar profiles, suggest personalised engagement strategies and even activate targeted upselling and cross-selling campaigns without human intervention. AI does not replace CRM systems; rather, it orchestrates and enhances them by integrating them with other enterprise systems — such as e-commerce, ERP and marketing automation — to create an intelligent, active and responsive ecosystem. 

This is why companies must rethink their technological infrastructure through an AI-ready lens: it is not simply about adding another tool, but about redesigning the way data flows, is enriched and ultimately used to generate value.  

The adoption of B2B e-commerce took years to consolidate in Italy. AI, by contrast, is dramatically accelerating timelines: what took decades to mature with e-commerce is now materialising in months with artificial intelligence. 

Precisely because of this acceleration, many companies are now facing a crucial question: does it make sense for us to adopt AI, and if so, how? There is still considerable uncertainty and difficulty in identifying real-world use cases that go beyond superficial experiments or now outdated applications, such as first-level chatbots limited to answering predefined FAQs. 

The key issue is the company’s level of digital maturity: it is impossible to make a technological leap toward artificial intelligence within organisations that are still poorly digitalised, characterised by fragmented manual processes and data scattered across silos or disconnected systems. 

AI requires solid foundations: clear business logic, a data-oriented mindset among people, structured information governance and systems capable of communicating with one another. Without these foundations, AI risks amplifying inefficiencies rather than solving them.