Case Studies & Insights

Technical breakdowns of real projects — architecture decisions, engineering trade-offs, and business impact. Published on LinkedIn.

Filter:
GEO · AI SEO · Website Audit 8 min read

Why Your Website Is Invisible to ChatGPT, Claude and Perplexity

You rank on Google but AI search engines skip your site entirely. Here is why: AI crawlers cannot execute JavaScript, and most business websites fail every other AI readiness check too. A practical guide to generative engine optimization (GEO): SSR, llms.txt, robots.txt, sitemap.

GEO AI SEO llms.txt SSR Website Audit
Full-Stack · AI Music · SaaS 5 min read

AI Music SaaS: Personalized Songs in Polish — hitdlaciebie.pl

Built a full-stack AI music SaaS for a client in two weeks with zero prior experience in AI music generation. Two-model lyrics pipeline (Gemini 2.5 Flash + Claude Sonnet 4.6), four music APIs tested, Suno V5.5 shipped. Real problems: SSRF, duplicate webhooks, Shopify test mode.

Next.js 14 TypeScript Suno V5.5 Shopify PostgreSQL
RAG · AI Agent · E-commerce 7 min read

Web Sales Assistant for E-commerce: da-vinchi.pl

A full-stack AI Sales Assistant for a Polish wallpaper store. LangGraph ReAct agent with 5 tools, Hybrid RAG (SQL + pgvector), ETL pipelines, buyer-scenario tests, and a web chat widget — production-grade, not a demo.

Python LangGraph FastAPI Supabase pgvector LangSmith
Automation · LLM Pipeline 6 min read

Invoice Automation Pipeline: IMAP → PDF → Gemini → Drive + Supabase

Invoices from multiple suppliers arrive as PDFs by email in different formats. Manual routine: find, classify, extract, file, enter into accounting. Now it runs on a schedule — automatically, end-to-end.

TypeScript Gemini 2.5 Flash GitHub Actions Supabase Google Drive
Full-Stack · Multimodal AI 4 min read

AI Image Composer: Identity Replacement in 30 Seconds

A Polish creative studio was burning hours on trial-and-error prompting for personalized AI paintings. I built a production web app in one day using React 19 and Gemini that turns it into a 30-second workflow.

React 19 TypeScript Gemini Canvas API Railway
Automation · B2B · Lead Gen 6 min read

B2B Lead Pipeline: 107 German Companies, $4.80 Cost, 22 Min

Built an automated B2B lead generation pipeline for the German market: Google Maps → Impressum scraping → Gemini enrichment → Supabase → CSV. 107 companies, 54 emails, §7 UWG compliant — all in 22 minutes for under $5.

Python Apify Tavily Gemini Supabase
Automation · AI Image · Retail 5 min read

Wallpaper Visualization Automation: AI Replaces Photoshop

Automated the manual Photoshop workflow for a wallpaper retailer: geometric transformation vs. AI generation — the architecture decision that matters. What works for Tapetenhändler, Küchenstudios, Möbelhäuser, and any German retail shop with room visualization needs.

Python FastAPI Gemini React Supabase
RAG · Architecture 4 min read

From Data Mirroring to Knowledge Synthesis: RAG Architecture Evolution

My first mistake was vectorizing raw PDFs. The result was a "calculator without a brain." Here is the architectural shift — Parent Document Retrieval, LCEL chains, and two-layer storage — that actually eliminated hallucinations.

LangChain LCEL pgvector Supabase RAG patterns
Automation · QA 5 min read

QA Automation for RAG Systems: Testing AI Pipelines

Building a RAG pipeline is one thing. Knowing it works reliably under edge cases is another. I built a testing framework with buyer-scenario integration tests, 98% data optimisation, and GitHub Actions automation.

Python LangGraph pytest GitHub Actions Supabase
RAG · AI Agents 5 min read

Vibe Coding a Hybrid RAG Agent: Supabase + Gemini File Search

A 57 MB PDF catalog, 55K+ tokens, ~$45/mo in hot-cache costs. I engineered a Hybrid RAG Telegram Sales Agent combining Supabase pgvector for hard data with Gemini File Search for semantic context — sub-second responses, near-zero API cost.

Hybrid RAG Supabase pgvector Telegram Gemini
💡

Common thread across all these projects

The real engineering is not in the LLM call. It is in what surrounds it: controlled ingestion, deterministic fallbacks, idempotent pipelines, and structured storage. That combination is what separates production automation from demos.

Impressum · Datenschutzerklärung · AGB · Cookie-Einstellungen