Javier Manzano
CEO & Founder at Soamee
Software engineer with over 10 years of experience leading web development, mobile app and artificial intelligence projects. Founded Soamee in 2017 from Madrid, focused on delivering high-quality artisan software for startups, SMEs and enterprises across Europe and Latin America. He is studying Philosophy and leads applied AI ethics at Soamee: turning principles into tests on real systems.
Experience
Over a decade building software for startups, SMEs and enterprises. Specialized in React, Node.js, Python and cloud architectures (AWS).
Specialization
Artificial intelligence, AI agents, LLM integration (GPT-4, Claude), web and mobile development, and scalable software architecture.
Global reach
Projects across Spain, Europe and Latin America. Working in three languages: Spanish, English and Portuguese. Clients in fintech, retail, healthtech and edtech.
Latest articles
AI literacy: what Article 4 of the AI Act requires after the Omnibus
Since 2025, Article 4 of the AI Act requires training for staff who use AI, and the Omnibus softened it. What it asks now, plus a role-based plan for SMEs.
llms.txt: What It Is, How to Create It, What to Expect
What llms.txt is, its format, how it differs from llms-full.txt, how we generate it on every build at soamee.com, and the mistakes that make it useless.
Labelling AI content: what the EU AI Act requires
Which AI-generated content must be labelled under the EU AI Act, what falls outside it, and how it applies to images, chatbots and marketing.
Typed AI vs generative: when to use each
LLMs generate free text. Jev returns typed values with probability. For extraction, classification or scoring, the difference matters.
Flux vs Midjourney vs DALL-E for business
A practical comparison of Flux, Midjourney and DALL-E for marketing, ecommerce and brand teams.
Bright Data in a scraper: when to use Web Unlocker, Scraping Browser or residential proxies
Which Bright Data product solves each block (403, CAPTCHA, 429, geo), how to connect it from Node.js and Playwright, and how to enable it per source.
We dropped Jira: AI agents and a tasks.md that fills itself in
With agents, we stopped needing Jira. We track work in a tasks.md the agents keep up to date. What we gained, what we lost, and what is still unsolved.
AI Model Map: October 2026
Claude Opus 5.5, GPT-6 Sol and Luna, Grok 4.7, Jev, Gemini. Where each model stands, what it does and which to choose for your use case.
How to audit chatbot bias with counterfactual tests
How to measure whether your chatbot treats two users differently when only their name, age or origin changes: test pairs, a blind LLM judge and statistics.
72 hours of vulnerabilities: what to patch, what to rotate and which certificates to replace
PHP, Node.js, GitLab, LiteLLM, Railway and 404,607 GoDaddy certificates between 19 and 22 October 2026. Fixed versions and steps, in order.
php2shell: the critical PHP vulnerability and what to do today
CVE-2026-109618 and CVE-2026-109619 allow unauthenticated code execution on PHP servers. How to check your version, update and mitigate.
Scraping in production: smoke tests, alerts and sources that change without warning
How we monitor 35+ scraping sources: live-site tests every three days, retry queues, alerts with a diagnosis and the silent zero-results failure.
Tell us your challenge. We'll propose a solution.
No commitment. We reply within 24 hours, and within 48 hours you have a proposal with scope, timeline and budget. No fine print.