I know it's January 11, 2026, and I'm only now wrapping up 2025. This delay isn't accidental - the following paragraphs will explain why the last months of the past year required my complete immersion in work that temporarily pushed aside even such important rituals as the annual summary.
When I look at 2025, I see the most intensive educational and technological period in my entire career. It was the year I crossed the boundary between being a passionate knowledge sharer and becoming an architect of real AI solutions.
"2025 taught me that true AI democratization requires more than education - it demands infrastructure, tools, and the courage to experiment with new forms of knowledge transfer."
Journey Through Three Training Programs
January brought an unexpected opportunity - participation in Google's "AI Skills of Tomorrow" program. Five weeks of intensive learning from personal productivity through AI marketing to organizational transformation. This program became my inspiration for effective use of tools like NotebookLM and ElevenLabs, which I had previously experimented with. I discovered the deeper potential of AI integration with Google Workspace.
Moreover, in the second edition of "AI Skills of Tomorrow 2," I returned not only as a participant but also as a Discord content moderator - supporting new participants became a natural extension of my mission to democratize AI knowledge. In February, I added the Google AI Essentials certificate on Coursera - a foundation that helped me understand the mechanisms of large language models.
June brought the AIDEAS program - a true journey from theory to practice. The culmination was creating a team of three collaborating AI agents for Saint-Gobain: Orchestrator Agent with Gemini 2.0, Product Agent with GPT-4o, and Analytics-Feedback Agent with Gemini 2.5 Pro. For the first time, I designed a system where different AI models work together, each with its own specialization.
Simultaneously, from June, the AI Heroes program from Microsoft was running - a fascinating exploration of the Azure, Copilot, and Power Platform ecosystem. Discovering that Google and Microsoft have fundamentally different AI philosophies enriched my toolkit as an educator: Google democratizes access, Microsoft integrates with existing workflows.
A collage of earned certificates and badges from 2025 - Google AI Skills of Tomorrow, Google AI Essentials, AIDEAS, AI Heroes, Microsoft Learn.
"Discovering different AI philosophies - from access democratization to deep integration - enriched my educator's toolkit. Now I can show not only 'how' but also 'why' to choose specific tools."
29 Articles - Knowledge Map
Throughout the year, I systematically published on my blog. From fundamentals (AI history, Moravec's paradox, Turing Test) through practical workshops ("Prompt Master" series, prompting techniques) to industry applications (AI in law firms, KIRP recommendations). Collaboration with UEK on the AI BOOST project showed how young people create impressive projects in just a few weeks - from sentiment analysis to AI-powered cookbooks.
I also created my first commercial products: the "Prompt Master in One Day" course and the "AI in Legal Practice" workshop with Codelaw.pl.
Social Media in Numbers
Bar or line chart showing LinkedIn post views in 2025 with marked peaks (May ~3400, July ~2300) and decline in Q4.
2025 brought significant growth in social media presence:
LinkedIn:
- 14,462 post views
- 1,085 new followers
- Building professional network with AI leaders and enthusiasts
Facebook, Instagram, Threads:
- 30,976 post views
- Systematic publication of educational content
- Building community around AI democratization
The LinkedIn views chart tells the story of the entire year: peak in May (~3,400 views), second in July (~2,300), and then... a drastic drop in Q4. This wasn't accidental.
"Sometimes the most important decision is to slow down publication pace and build something lasting. The visibility drop in Q4 hurt, but without this infrastructure, I couldn't scale the mission in 2026."
Silent Work at Year's End
Simplified infrastructure architecture diagram - 7 icons representing services (n8n, Baserow, Postiz, Twenty CRM, Qdrant, Vaultwarden, Uptime Kuma) connected by lines showing integration .
August-December was a period of "silent work." I made a decision that cost me social media visibility but built the foundation for the future. During the last months of the year, I took on the role of Senior DevOps Engineer, designing and implementing comprehensive Self-Hosted infrastructure for AI for Everyone.
VPS on OVH, Docker Stack with Zero Trust architecture, 7 integrated services: n8n (automation), Baserow (database), Postiz (social media), Twenty CRM, Qdrant (vector database), Vaultwarden (security), Uptime Kuma (monitoring). Implementation of RAG with Google Gemini, intelligent assistant with access to knowledge base.
Agency OS is also emerging - an operating system for tech agencies, which I'm building using Google Antygravity and the VibeCoding method. It's a fascinating experience showing that you can create advanced AI systems without being a programmer - all you need is a clear vision and the ability to communicate with AI. The system, where the AI Orchestrator will semantically analyze orders and autonomously assign tasks, is just being born, but its concept and first elements are already functioning. This is a project showing how far you can go with AI as a partner in creating solutions.
"You can create advanced AI systems without being a programmer - all you need is a clear vision and the ability to communicate with AI. This shows how far you can go with AI as a partner in creating solutions."
Biggest Lessons of the Year
Authenticity in Education - I cannot teach what I don't deeply understand myself. The six-month training marathon gave me not only theoretical knowledge but practical experience with a wide spectrum of solutions.
Value of "Silent Work" - sometimes the most important decision is to slow down publication pace and build something lasting. The visibility drop in Q4 hurt, but without this infrastructure, I couldn't scale the mission in 2026.
From Student to Architect - I traveled the path from enthusiast through certified specialist (Google, Microsoft, AIDEAS) to systems architect building real infrastructure.
What Will 2026 Bring?
Futuristic illustration presenting three pillars: 1) Robot/AI assistant at computer (automation), 2) Building/code (Agency OS under construction), 3) AI Avatar before group of students (training).
The built infrastructure will become the foundation for implementing three key plans:
Publication Automation with Human in the Loop - using my own infrastructure, mainly n8n, for partial automation of content creation process. It's not about replacing human creativity, but intelligent support: from research through drafts to distribution. I remain at the center of the process as curator and final decision-maker.
Continuation of Agency OS Work - further development of the system using Google Antygravity and VibeCoding. This is a long-term vision that will evolve with my growing experience.
Revolution in Training - launching the first online training sessions, both free (knowledge democratization) and commercial (project financing), conducted by AI avatars. This will be a breakthrough - the ability to scale education while maintaining quality and personalization.
2025 taught me that true AI democratization requires more than education - it demands infrastructure, tools, and the courage to experiment with new forms of knowledge transfer.
Starting 2026, I'm no longer just an AI educator - I'm a solutions architect who can show the full cycle: from idea through implementation to a working system. And this opens completely new possibilities for serving the AI for Everyone mission.
This article was created in collaboration with Claude Sonnet 4.5, showing how human-AI synergy works in practice - from concept to publication.
