RAG is the aspirin of generative AI: everyone prescribes it, almost no one truly understands how it works, and when it fails, the patient hallucinates. After implementing RAG in production with the Ops Copilot and discarding it in favor of Tool Calling for the obsolescence radar, I’ve catalogued the 7 errors that destroy precision — and the solutions that actually work.
On August 2, 2026, most obligations of the EU AI Act come into force. Maximum fine: 35 million euros or 7% of your global turnover. And most articles explaining this regulation are written by lawyers, for lawyers. Not this one. This one is written by an engineer who already operates AI agents in production, translating every article of the regulation into the language we actually understand: architectures, pipelines, and code.
Today, every LLM has its own way of connecting to your databases and tools. It’s a chaos of proprietary integrations. MCP (Model Context Protocol) promises to end this by becoming the ‘universal plug’ of AI. We analyze if it’s truly the future or just another standard that will die trying.
Everyone talks about what AI can do. Almost no one talks about what AI costs to do. After more than 6 months operating autonomous agents in production with Gemini, CrewAI and GitHub Actions, I open the books and break down every euro invested — and every surprise I found along the way.
Transforming the Obsolescence radar into an Event-Driven 24/7 unified microservice. We cover dual ingestion vectors (Accuris vs. Raw Emails), non-blocking HTTP processing through BackgroundTasks, and closing the operational loop via Microsoft Teams.
Destroying the ‘Chat with PDF’ myth: How to orchestrate Gemini 2.5 via CrewAI so it delegates deep mathematical responsibilities to a pure relational tool connected to Supabase, hitting 0% hallucination on P&L impact calculations.
1. The Execution Chasm Companies invest millions in planning software and ERPs. They spend months integrating data, cleaning records, and tuning demand forecasts. Yet, at the end of the day, execution often relies on an analyst staring at a dashboard. A human reading charts, finally drafting an email to the plant director to request production adjustments. That human delay is inefficient, error-prone, and, frankly, unnecessary in the modern era. Pure mathematics has no value if it isn’t executed. ...
When visitors have questions about my 72 blog posts, they shouldn’t have to dig through archives. I built an AI copilot that uses Algolia Agent Studio to deliver conversational answers backed by real content.
While the world awaits the arrival of fully functional quantum computers, a company from San Sebastián is using the mathematics that underpin them to solve AI’s most urgent problem today: efficiency. Multiverse Computing has developed a technology based on ‘Tensor Networks’ that compresses large language models (LLMs) by up to 95%, allowing them to run on standard hardware and edge devices. This pragmatic ‘quantum-inspired’ approach has earned them a $500M valuation and partnerships with giants like Bosch and Iberdrola. We analyze their strategy, technology, and business model.
Remember your first day amidst mountains of paperwork? Traditional onboarding costs time and money, and often fails in its mission. A recent report reveals alarming figures: 24 days on average to complete the process and only 12% of employees satisfied. The solution is not more paper, it’s Artificial Intelligence. Discover how the combination of IDP for bureaucracy and GenAI as a virtual mentor can automate document management and personalize information delivery, freeing up HR to focus on culture and people.