IMPLEMENTATION NOTES
Implementation notes: what remains after day five
Every implementation leaves a trace: an architectural decision, a bug that cost half a day, and a solution that survived contact with a real user. We write it down here — before it evaporates. No trend reports, no corporate conclusions. Just what we built and what we learned.
TECHNICAL SEO · 6 MIN READ
schema.org structured data: what to actually implement on a small business website
Structured data is not ranking magic. It is a way to stop machines having to guess who you are and what you sell.
Read articleGEO · 5 MIN READ
llms.txt: a guide to your company for language models
One text file where you tell the machine directly: this is what we do, this is how we work, this is how to cite us.
Read articleLOCAL SEO · 6 MIN READ
Local SEO: how a small service business can win in its city
You don’t have to beat the entire internet. You just need to beat five businesses within a ten-kilometer radius.
Read articlePERFORMANCE · 5 MIN READ
Mobile page speed: what really changes visibility and conversion
Most small business websites aren’t slow because of their framework. They’re slow because of three images and five scripts.
Read articleGENERATIVE EO · 6 MIN READ
How to get ChatGPT and Perplexity to cite your business
A model doesn’t “like” brands. It takes the passage that can be copied in full and sounds like a fact, not an ad.
Read articleSEO · 6 MIN READ
Service pages that sell: one service, one page, one intent
A “Services” page listing eight services ranks for nothing. Eight separate pages rank for eight things.
Read articleAI-FIRST · PROCESS · 7 MIN READ
The AI-first pipeline: how a working application is built in 5 days
We don’t shorten projects by skipping stages. We shorten them because writing code is no longer the bottleneck.
Read articleAI IN BUSINESS · 7 MIN READ
Your own AI assistant using company documents: how RAG works and when it makes sense
A chatbot that knows your price list is useful. A chatbot that makes it up is costly.
Read articleAUTOMATYZACJA · 6 MIN READ
Invoices, statements, PDFs: how to turn documents into data you can trust
The model will read 99% of line items. All the engineering work is about knowing which items make up that one percent.
Read articleAUTOMATYZACJA · 6 MIN READ
AI agents in a small business: what really works, and what is just a conference demo
An agent handles messy input brilliantly. It cannot handle the lack of a decision about who is responsible for the outcome.
Read articleARCHITEKTURA · 9 MIN READ
Multi-tenancy in Postgres: how we built a system for a workshop network where branch data never gets mixed up
One system, eleven branches, one database. The hardest part was not building features — it was proving that the Gdańsk branch would never see a job from Kraków.
Read articleSHORT ANSWER
Mocne.ai implementation notes are practical records from real AI-First projects: LLM agents, RAG systems, document parsing, process automation for SMEs and generative engine optimisation (GEO). Every post includes code, architecture and verifiable facts.
- We only write about what we have implemented for clients ourselves.
- Topics: LLMs, RAG, document parsing, AI agents, AI-First pipelines, GEO.
- Every note includes code, an architecture diagram or an action list.
- Two languages: Polish and English. Author: the Mocne.ai team.