Glossary
AI and Search Glossary.
Plain-English definitions of the AI, search and software terms we use, each checked against primary sources such as Google Search Central and schema.org. Each entry starts with a short answer, then says where we use the term and which sources back it.
A
- Agentic AI
- Agentic AI is a broad label for AI systems that can work toward a goal with limited supervision. It usually means one or more AI agents that plan steps and use tools. Sometimes several agents work together.
- AI agent
- An AI agent is a program that uses an AI model to choose its own next step toward a goal, such as checking a calendar or drafting a text. It works inside limits that people set.
- AI citation
- An AI citation is a link or source name that an AI tool shows with its answer to back up what it says. Perplexity shows numbered links, and Google’s AI Overviews show links to supporting websites. Being cited is not the same as getting a click.
- AI implementation framework
- An AI implementation framework is the order of work once the job already has a name. Name the job, score the data, write the process, assign a checker, then keep the work or stop it.
- AI Overviews
- AI Overviews are AI-written summaries Google can show in Search, with links to the pages they used. A page must be indexed and allowed to show a snippet to be one of those links.
- AI pilot
- An AI pilot is a small trial of an AI tool on one named job. It comes before a wider rollout. It stalls when the demo works and the daily work does not.
- AI readiness
- AI readiness is how prepared a business is to put an AI tool to work and get value from it. It depends on the business more than the tool. That means its data, the way work gets done, its people and who makes decisions.
- AI readiness assessment
- An AI readiness assessment asks whether one business can put a new tool to work. It looks at the data, the process, the people, and the decision.
- AI readiness framework
- An AI readiness framework scores the business before you score a tool. We look at data, process, people, and the decision. Each group gets a word you can defend.
- AI visibility (AI search visibility)
- AI visibility is how often a business shows up in answers that AI tools write, and how well. It might be named, quoted or linked. There is no standard way to measure it, so each tool and report counts it in its own way.
- AI-ready data
- AI-ready data is one current place the facts for a job live. A system can read it. A person knows which fields look fine but are not.
- Answer engine
- An answer engine is a search tool that replies to a question with a written answer, usually with links to its sources, instead of only a list of links. Perplexity calls itself one. Google’s AI Overviews and AI Mode work in a similar way.
- Answer engine optimization (AEO)
- Answer engine optimization (AEO) is the work of getting your pages used and cited when an AI tool answers a question directly. Google says it is still SEO.
B
- Business process automation (BPA)
- Business process automation (BPA) means using software to do the repeated steps of a business job, like sending a bill to be approved. The aim is less manual work, fewer errors and the same result each time. It can run on simple rules, workflows or AI.
C
- Crawler (bot)
- A crawler is a program that visits web pages on its own and reads them, usually for a search engine or an AI company. It is also called a robot or spider. Sites use a robots.txt file to tell crawlers which pages they may visit.
- Custom software
- Custom software is software built for one business or group of users, around how they already work. Off-the-shelf software is different: it is made for many buyers and sold as a ready product. Custom software makes sense when no ready product does the job well.
G
- Generative AI
- Generative AI is AI that creates new content, such as text, images, audio, video or software code, in response to a request. Large language models, which write text, are the best-known kind.
- Generative engine optimization (GEO)
- Generative engine optimization (GEO) is the work of shaping web pages so AI tools that write answers are more likely to show and cite them. Examples of these tools are Google’s AI Overviews and Perplexity. Google says this work is still SEO.
- Gradient boosting
- Gradient boosting is a machine learning method that builds one strong model out of many weak ones. The weak ones are usually small decision trees. Each new tree learns from the errors the model still makes. It is added in a small step to shrink them.
H
- HQ
- HQ is Abracadabra’s client workspace, where our team and a crew of digital workers get your work done. Each request becomes a card with a plan. Nothing reaches your folders or your social networks until it has been reviewed and you have approved it.
L
- Large language model (LLM)
- A large language model (LLM) is an AI model trained on a huge amount of text. It writes by predicting the next word, or part of a word, over and over. Google’s Gemini and OpenAI’s GPT models are examples.
- llms.txt
- llms.txt is a proposed file, written in Markdown, that gives AI tools a short summary of a website and links to its most useful pages. It usually sits at the site’s root, at /llms.txt. Google Search says it does not use the file.
M
- Machine learning
- Machine learning is a way to build software that learns patterns from past data. It does not rely only on rules a person wrote. The trained model can then make predictions about new data, such as next month’s orders.
- Model Context Protocol (MCP)
- The Model Context Protocol, or MCP, is an open standard for connecting an AI app to outside systems. One connector can reach a file, a database, or a tool.
R
- Readiness Check
- The Readiness Check is our online check for a business that wants to put AI to work. It takes five to seven minutes. It scores readiness, growth and website visibility, and names the stage your business is at. It gives you next steps for your business.
- robots.txt
- A robots.txt file sits at the root of a site and tells crawlers which URLs they may fetch. Google uses it to manage crawl traffic. It is not a lock on a page.
S
- Schema markup and JSON-LD
- Schema markup is code added to a web page that labels what the page is about, such as an article, a product or a business. The labels come from a shared list called schema.org. JSON-LD is the format Google recommends for it. It is a block of code that machines read and visitors do not see.
- Search engine optimization (SEO)
- Search engine optimization, or SEO, helps search engines understand your pages. It also helps people find your site and decide to visit.
- Search impressions
- A search impression is one time a person saw a link to your site on Google. It counts the view, not the click. Search Console keeps it apart from AI-feature impressions.
W
- Workflow
- A workflow is the set of steps that takes a piece of work from start to finish, in a set order. In AI, it also means a system where an AI model and tools follow steps written in code ahead of time. That is different from an AI agent, which picks its own next step.