Symfony AI is getting a Core Team: what this changes for an AI stack in PHP
The creation of a Core Team confirms Symfony AI's ambition. It improves the project's readability, without removing issues of stability, observability, or vendor dependency.
PartITech blog
Technology blog
Technical analysis, practical guides and field experience to build, modernize and maintain sustainable applications.
Symfony Language Tools: what the official LSP changes for PHP teams
The new official LSP server includes routes, services, Twig, Messenger, and Doctrine. An important advancement, to be tested with the same requirements as a tool running the application locally.
WordPress 7.1 “Mary Lou”: what to test before updating
WordPress 7.1 brings visible changes to the editor, media and blocks, but also several technical changes which require thorough acceptance testing before deployment.
Agency commerce and UCP: preparing your catalog and checkout for AI-driven purchases
Agency commerce is not limited to displaying a product flow. To allow an agent to compare and then purchase, the merchant must make each step explicit, verifiable, idempotent, and reversible.
Digital accessibility: applying RGAA 4.1.2 today and preparing RGAA 5
Accessibility is not limited to an automatic score or a final audit. It must be designed, tested, and governed throughout the entire lifecycle.
In-depth research agents: building a chain of evidence rather than a convincing report
A clickable citation is not a guarantee of truth. Reliable research links every important claim to a source that has actually been consulted, keeps contradictions, and indicates what remains uncertain.
Mistral AI in 2026: how to compose an enterprise stack between open models, agents, OCR, and European infrastructure
Mistral no longer offers just a generation API. In 2026, the ecosystem covers open models, agents, document intelligence, search, connectors, customization, and infrastructure options. The challenge is to choose an architecture rather than a product catalog.
WordPress, Drupal, Symfony or headless: which architecture for a complex web project?
The right choice does not oppose a 'small CMS' to a 'large framework.' It consists in placing the editorial, the business, and the integrations in the components that serve them best.
AI Act applicable since August 2, 2026: the concrete action plan for a French company
Since August 2, 2026, the application of the European AI regulation has reached a major milestone. The first urgency is not to produce a hundred documents, but to know which systems exist, what role the company plays, and what evidence is actually expected.
Shadow AI: regaining control without blocking useful uses of generative AI
Banning all assistants does not eliminate usage: it makes it invisible. An effective strategy combines visibility, simple rules, an approved environment, technical controls, and quick handling of business needs.
Evaluating an LLM or an AI assistant: quality, latency, cost, and regressions
A public benchmark ranks models on general tasks. An evaluation produces a measurement of your quality, with your data, your constraints, and your costly errors.
GDPR and generative AI: mapping data before connecting a model to the company
A prompt is not the only place where personal data can appear. Indexed documents, logs, traces, memory, outputs, feedback, and connected tools form a chain that must be made visible before choosing the safeguards.