AI & tax glossary

Short, plain-language definitions of the AI and tax terms used across the MiCode blog — no need to hunt through a specific article to find them.

AEO

AEO (Answer Engine Optimization) is writing and structuring content so AI assistants like ChatGPT quote or cite it directly, the way SEO does for search engines.

AI Agent

An AI agent does more than answer: it plans the steps and uses tools on its own — querying systems, reading documents, then carrying the task out.

API

An API is the interface two systems use to exchange data automatically, with no Excel exports and no copying by hand.

ARP

ARP (the Industrial Development Agency) is Poland's state agency for industry, running digitalisation grants for manufacturers among other things.

Cache

A cache reuses the repeated, unchanging part of a prompt, which model providers charge less for — so a well-ordered prompt can cost several times less.

Chunk

A chunk is one of the fragments a document is split into, so an AI can retrieve and quote the relevant piece instead of reading the whole file.

CTR

CTR (Click-Through Rate) is the share of people who saw a link and clicked it — a measure of whether content invites action, not just impressions.

EDIH

An EDIH (European Digital Innovation Hub) is an EU co-funded centre where a company can test AI solutions and get advice for free or for a small fee.

Embedding

An embedding turns the meaning of a text into numbers, which lets an AI find passages that mean something similar even when they use different words.

ERP

ERP (Enterprise Resource Planning) is a company's central system holding orders, inventory, accounting and HR in one place instead of several separate tools.

FA(3)

FA(3) is the current version of the schema an invoice has to follow for KSeF to accept it, and a change of schema version forces an update in a company's own systems.

FENG

FENG (European Funds for a Modern Economy) is Poland's main national programme funding company innovation and R&D in the 2021–2027 period.

GEO

GEO (Generative Engine Optimization) is being present inside AI-generated answers themselves, rather than in the list of links beneath them.

GPAI

GPAI (general-purpose AI) means a general-purpose model such as GPT; the AI Act puts the documentation duties on its provider, not on a company building on top of it.

GPT

GPT is a family of large language models built by OpenAI (the technology behind ChatGPT); a version number like GPT-4 marks a specific generation of it.

Halucynacja

A hallucination is an AI answer that sounds convincing but is simply untrue: the model fills a gap with a guess instead of admitting it does not know.

KPO

KPO (Poland's National Recovery Plan) is how the country spends its share of the EU recovery fund; in practice its calls are closed to new applications.

KSeF

KSeF is Poland's mandatory national system for issuing and receiving e-invoices directly with the tax authority.

LangGraph

LangGraph is a developer framework for building AI agents that follow multi-step workflows with branching logic, rather than answering in a single pass.

LLM

An LLM (Large Language Model) is the type of AI model — like the ones behind ChatGPT — trained on huge amounts of text to understand and generate human language.

MŚP

SME (small and medium-sized enterprise) is the EU category for companies with fewer than 250 employees, and belonging to it decides access to most grants.

NCBR

NCBR (the National Centre for Research and Development) funds R&D projects, and its grants require genuine research work rather than simply buying tools.

OCR

OCR (Optical Character Recognition) turns a scan or photo of a document into text a computer can read — the first step to an emailed invoice entering your system by itself.

Orkiestracja

Orchestration is coordinating several models, tools and agents so they work on one task in a defined order and hand results to each other.

PARP

PARP (the Polish Agency for Enterprise Development) is the institution running most grant calls for small and medium-sized companies.

Pipeline

A pipeline is a fixed, predefined sequence of processing steps — unlike an agent it does not choose the order, it always runs the same one.

Prompt

A prompt is the instruction you give an AI model, and how precisely it is written decides whether the answer comes back useful or vague.

RAG

RAG (Retrieval-Augmented Generation) lets an AI look up your own documents before answering, so its replies are grounded in your data instead of guesswork.

SaaS

SaaS (Software as a Service) is software you use over the internet for a subscription fee, instead of installing and maintaining it yourself.

SEO

SEO (Search Engine Optimization) is preparing a site so it ranks high in search results.

SQL

SQL is the language database questions are written in — it is what "show last quarter's sales" looks like once a database can answer it.

Text2SQL

Text2SQL lets someone type a plain-language question and have it automatically turned into a database query, no SQL knowledge required.

Token

A token is a fragment of text, usually a piece of a word, and it is the unit AI models measure length and charge by — bills are counted in tokens, not in questions.

UPO

UPO is the official confirmation that the tax system accepted a document; only that receipt, not the act of sending, means an invoice exists in KSeF.