DIGITAL TECHNOLOGY IN PRACTICAL TERMS

Ecommerce Manager, a developer at heart.

I manage e-commerce projects on a day-to-day basis, and I code the tools that go with them. Here, I document the things that really interest me: cybersecurity, automation, AI and data.

See the blog About me

4 angles of attack

e-commerce, development, IA, automatisation

Areas of expertise

E-commerce

Développement Front-end

Cybersécurité

Automatisation & IA

Latest articles

See all →

Latest articleCybersécurité

MAI-Cyber-1-Flash: Microsoft Bets on AI to Industrialize Cybersecurity

Microsoft has launched MAI-Cyber-1-Flash, its first cybersecurity-focused AI model, integrated into MDASH. By pairing this compact model (handling 90% of tasks) with heavier models for edge cases, Microsoft claims a 96% score on the CyberGym benchmark while cutting costs by 50%. The announcement also introduces Perception, an agentic system for continuous threat monitoring. A concrete illustration of smart model-routing based on task complexity.

Read article →

E-commerceNew

AI Overviews in France: What It Really Means for Advertisers

AI Overviews and AI Mode rolled out in France in late July 2026, but ads embedded directly inside these AI blocks aren't live there yet — only classic Search and Shopping ads appear around them. CPCs are rising and clicks are getting scarcer, pushing advertisers to target higher up the funnel instead of focusing only on bottom-funnel, transactional keywords. Automation (PMax, AI Max) should be adopted gradually, not as a full replacement for traditional Search — going all-in too fast risks losing control over performance. Tracking and data quality (a clean CRM, Consent Mode V2, server-side tracking) are becoming prerequisites for Smart Bidding to work properly. Ad performance increasingly depends on overall marketing strategy: brand awareness, content, and multichannel presence.

Automatisation & IANew

ATLAS: what Google just taught us about how people really use AI

Google has launched ATLAS, the first large-scale study of how people really use its AI tools (Gemini, AI Mode, API), based on 15 million anonymized interactions across 150 countries. Key finding: AI is widely adopted at work, but used narrowly — around 21% of tasks in a typical job — and mostly for collaboration (ideation, research, learning) rather than automation. It's also reaching manual and technical trades, and creating significant value outside of work (admin tasks, purchases) that's often invisible in standard economic stats. Global adoption broadly tracks national wealth, with a few notable exceptions.

E-commerceNew

AI Overviews Land in France: What It Means for Your Product Pages

Google launched AI Overviews in France on 22 July 2026. Drawing on two years of feedback from the US, this article analyses the actual impact on e-commerce traffic: product pages are largely unaffected, but all informational content — buying guides, comparison features and blog posts — is on the front line