35,000 API Calls to Say 'Nothing to Learn Here'
A reflection hook in my multi-agent system fired 35,000 LLM calls in three days. Cost: $165. Useful output: zero.
Posts about ai
Are AI Tools Worth Paying For? What the Data Actually Says
1B transactions: AI apps convert 52% better but retain 30% worse. Is ChatGPT Plus or Claude Pro worth paying for?
Vibe Coding: What It Is and Why 90% Do It Wrong
Vibe coding is real and growing. It also has serious problems. No-hype guide: tools, mistakes and how to use it right.
Will AI Replace My Job? What the Data Says
Anthropic published the most serious study on AI and employment. Nuanced data that goes against both narratives.
Coin Puzzle: 0 or 1/13? Why AI Picks the Wrong One
3 coins, P(heads)=1/3, tails always even: is the answer 0 or 1/13? Full math and why this puzzle breaks every AI.
How to Build AI Agents for Free (No-Code and Code Options)
Free tools for AI agents: no-code, low-code and open source. What to use for your case and what free really means.
n8n AI Agents: What You Can Actually Automate
n8n has native AI nodes, memory, and tool calling. What works well, where the real limitations are, and when it makes sense for agents.
Why AI Agents Fail in Production (And Nobody Tells You)
AI agents work in demos. They break in production. Six failure modes nobody explains — and what actually helps.
AI and the Pentagon: Silicon Valley's New Battleground
OpenAI, Google, Meta and Anthropic compete for military contracts. Silicon Valley's pivot to the Pentagon is real.
AI Weekly Radar: February 10-18, 2026
Sonnet 4.6 redefines AI pricing, Mistral acquires Koyeb, EU blocks chatbots, India attracts $200B, and 17 startups raise $100M+.
Claude Sonnet 4.6: Flagship Performance at Mid-Tier Pricing
Sonnet 4.6 matches or beats Opus 4.6 on several benchmarks at 40% less cost. Frontier intelligence is no longer expensive. What it means for your AI budget.
FinOps for AI: How to Stop Bleeding Money on Inference Costs
Your AI pilot cost €50/month. Production is €5,000. A practical FinOps guide to controlling inference costs, tokens, embeddings, and LLM API spend in 2026.
Claude Opus 4.6: the model that tanked the market
Claude Opus 4.6 brings 1M context and agent teams. Cowork plugins wiped $285B off software stocks. What changed and why it matters.
Enterprise AI in 2026: Where We Actually Stand
84% of companies haven't redesigned a single role for AI. We analyze Deloitte's 2026 report and what businesses must do next.
LLM Sycophancy: AIs That Only Tell You What You Want
Stanford identifies artificial flattery as one of AI's major problems in 2026. If you've used ChatGPT lately, you know exactly what they're talking about.
AI Maturity: Less Code, More Focus
Mature AI means less code, not more. The shift isn't smarter tools — it's knowing when to stop engineering and focus.
The Hardware Crisis: Why Your Next Laptop Will Cost 20% More
RAM +344%, SSDs +100%, laptops +20%. AI data centers are devouring the world's chip production. Welcome to the hidden cost of the AI revolution.
OpenClaw: The Viral AI Assistant I'm Not Installing (Yet)
OpenClaw: 15K GitHub stars. An AI agent managing your PC sounds great... until you see the permissions it needs.
Only 5% See Real ROI from AI: the Uncomfortable Truth
70-80% of agentic AI projects die before production. Real cases from Equinor ($330M saved) and Travelers (20,000 users, 50% claims automated).
BigQuery Now Talks Back: Google Makes GenBI a Reality
Google turns BigQuery conversational with GenBI. What changes, what doesn't, and why it won't replace the analyst.
AI that works: Data-ka and automatic tender proposals
A Basque company uses NLP to read 200-page tender documents and generate proposal drafts. No hype, no smoke, just measurable results.
Weekly AI Radar: January 22-29, 2026
DeepSeek shakes the market, OpenAI and Anthropic compete for healthcare, 95% of GenAI projects don't generate ROI, and Spain gets an AI factory in Barcelona.
ChatGPT drinks from Elon Musk's fountain
OpenAI integrates X/Twitter data into ChatGPT. What it means when your chatbot drinks from Musk's fountain.
OpenAI Prism: the end of the generic chatbot
A free workspace with GPT-5.2 for writing papers in LaTeX. But the real news isn't Prism: it's what it represents about where AI is heading.
Synthetic Data: The $8B Business of Making Up Data
Nvidia just paid $320M for Gretel Labs. The synthetic data market is exploding. What it is, why it matters, and why you should care.
On-premise is back: why companies are fleeing AI cloud
Companies that went all-in on AI cloud are pulling back. Latency, costs and vendor lock-in force a rethink.
AI Slop: why we're sick of AI-generated content
The internet is filling up with mass-produced AI garbage. What AI Slop is, why it's a problem, and how to tell real content from noise.
GenBI: the end of the data analyst bottleneck
GenBI: talk to your data in natural language. What works, where the limits are, and why it won't replace analysts.
Implementing AI in small business: the real truth
The truth about AI projects in SMBs. What works, what doesn't, actual costs, and how to avoid wasting money.
Cursor vs Windsurf vs GitHub Copilot: honest comparison 2026
Real comparison of the 3 most popular AI-powered IDEs. Pricing, models, workflow and which to choose based on your profile. Tested on real projects.
Data-Centric AI: why more data doesn't mean better models
The paradigm shift in machine learning: invest in data quality, not bigger models. Tools and practices to implement it.
DeepSeek: cost, performance and data sovereignty
Low costs, high performance, open source... and your data on Chinese servers. Risk analysis and alternatives for Western companies.
How AI thinks: System 1 and System 2 in LLMs
Do language models reason or just improvise? The answer through Kahneman's framework and what it means for how we use them.
My AI development stack in 2026: what I use and why
The AI tools I use daily for coding, automation and productivity. No sponsorships, just what actually works.
The copilot lie: why 95% of companies aren't seeing results
Studies from METR, MIT, and CMR show AI isn't delivering promised productivity gains. What successful companies do differently.
17% of Basque Companies Use AI — and They Earn 8.7% More
While 95% of AI pilots fail globally, the Basque Country shows a model that actually works. Analysis of the BAIC 2025 diagnosis.
AI Datacenters: 4% of US Electricity — Nuclear Is Back
AI's energy demands are resurrecting Three Mile Island. Analysis of the nuclear renaissance driven by Big Tech and its implications.
$25B Patagonia datacenter: opportunity or extractivism?
$25B datacenter in Patagonia backed by the US. Record investment or digital extractivism? What Argentina gives up.
Cursor AI: Complete Guide to the AI-Powered IDE
What Cursor is, how it works, when to use it vs Copilot, updated pricing, and tips to get value from day one.
"That's impossible" — ignore whoever says it
What AI can actually do for your business. Real cases, realistic expectations, and why those who say "impossible" are usually the ones who don't know how.
Generative AI: What It Is and How It Works
What is generative artificial intelligence, how it works, real examples, and why it's changing everything. No jargon.
AI Agents for Enterprise: From Demo to ROI in 2026
AI agents for business: real examples (CaixaBank, Salesforce, EY), what works, what doesn't, and how to move from pilots to production with measurable ROI.
AI Trends in 2026: What Actually Matters
The 7 AI trends defining 2026: autonomous agents, small models, MCP, ROI focus and more. A guide for data professionals.
The 7 Best Free AI Tools in 2026 (ChatGPT Alternatives)
You don't need to pay $20/month. These are the best free AIs for 2026: which to use for what and when paying is actually worth it.
ChatGPT vs Gemini vs Claude: 2026 Comparison
Practical guide to choosing between the big three LLMs. Pricing, capabilities, use cases, and which one fits your situation.
DeepSeek: what it is and how to use it for free
Complete DeepSeek guide for 2026. Free ChatGPT alternative: web, mobile app, and local installation with Ollama.
MCP (Model Context Protocol): What It Is and Why
Complete guide to MCP, the standard connecting AI agents to your systems. How it works, available servers, and practical examples.
Power BI Copilot in 2026: Cost, Setup, and Honest Review
Complete guide to Power BI Copilot: requirements, Fabric pricing, how to enable it, real use cases, and honest limitations.
Prompt Engineering Guide: How to Talk to LLMs
Everything you need to know to write effective prompts. From beginner to advanced, with practical examples and the limits nobody tells you about.
What Are AI Agents (explained without jargon)
Beginner's guide: what an AI agent is, how it differs from ChatGPT, real examples, and where to start.
World Models: AI that predicts the physical world
What World Models are, how V-JEPA works, and why Yann LeCun is betting $3.5 billion on them.
LeCun leaves Meta: LLMs won't reach human-level AI
The godfather of AI bets $3.5 billion on a different architecture: World Models.
The AI bubble: 7 trillion looking for returns
Who wins, who loses, and why you should care. Analysis of massive AI investment and its bubble signals.
95% see no results with AI (and why that's normal)
The J-curve of adoption nobody tells you about. Why productivity drops before it rises when you adopt AI.
90% of your data is garbage nobody knows how to process
Why companies buy AI without having their data ready. The data plumbing problem.
Don't be a fanboy of any model
A year of real data using AI to code. My Cursor usage graph and why I jumped from model to model.
Taxonomy of LLM failures
Language models fail in four distinct ways. Each requires a different fix: prompt tuning, RAG, fine-tuning, or guardrails. A practical taxonomy.
The model knows how to reason. It just won't commit
17 prompt iterations revealed that the model finds the correct answer but self-censors for not being standard
More tokens doesn't mean better results
How an exhaustive meta-prompt caused context overflow and reached the same error on a random walk problem
Prompt for ambiguous problems (3 coins puzzle)
3 coins, P(heads)=1/3, tails always even — 0 or 1/13? A prompt that forces LLMs to find the correct interpretation.
Why LLMs Reject Their Own Correct Answers
Two-Box separates contexts so the LLM reviews without bias. Problem: counterintuitive answers get discarded.
36% of Spanish companies now use AI: what it means for you
AI adoption data in Spain in 2025 and what it means for those who work with data. A view from Andalusia.
My Claude workflow: no lost context, free docs
My system for keeping Claude on track during long conversations and getting free documentation out of the process.
When AI 'fixes' your bug by hardcoding the print
The time Claude Sonnet 3.5 decided the best way to fix a bug was to lie straight to my face.
AI-generated code works, but is it secure?
An AI-generated docker-compose can work perfectly while having a massive security hole. Real case and checklist to avoid it.
You have 3-5 years before AI agents become normal
78% of executives believe digital ecosystems will be built for humans AND AI agents. The window to position yourself is closing.
Using AI cost $1000. Now it costs $1. What's your excuse?
AI costs dropped 1000x in two years. If you're not using it, it's not about money. It's fear, ignorance, or laziness.
AI is the new data leak channel (and nobody's ready)
Employees copy sensitive data into ChatGPT without thinking. AI-powered phishing is more sophisticated than ever. Traditional security doesn't work.
AI is running out of internet to eat
AI models consume data faster than we create it. Quality internet content is almost used up. What comes next?
AI won't take your job. Someone who uses it better will.
Excel has been around for 40 years and hardly anyone masters it. AI will be the same. Your edge isn't the tool—it's how long you've been using it.