AI terms

AI terms, explained for teams that don't code

Each page explains one AI term in plain words, with dated sources: what it means, why it matters to a small team, a worked example, and the first steps. Each ends with a free AI skill that puts the idea to work, which you can add to your team's pod and run in Claude, ChatGPT, or a coding agent.

New to all of this? Start with AI skills and MCP, the two that decide how your team shares its AI work. Or browse the AI skills in the library.

• What is RAG?RAG lets an AI look up your documents before it answers. What it is, where the term comes from, why a small team should care, a worked example, and how to start.
• What is GraphRAG?GraphRAG builds a knowledge graph from your documents so an AI can answer questions about the whole set. How it differs from RAG, when it's worth it, and an example.
• What is a knowledge graph?A knowledge graph stores facts as things and the relationships between them. Where the term comes from, why AI teams use it, an example, and how to build a first one.
• What is an ontology?An ontology is the agreed list of the kinds of things in your business and how they relate. What it is, why AI projects need one, an example, and how to draft one.
• What is context engineering?Context engineering is choosing everything an AI sees when it answers: instructions, examples, documents, tools and history. What it is, why it matters, and an example.
• What is MCP?MCP is the open standard that connects AI apps like Claude and ChatGPT to outside tools and data. What it is, who runs it, what to check before connecting a server, and an example.
• What is a forward deployed engineer?A forward deployed engineer works inside a customer's team to make software fit their real work. Where the role comes from, why AI startups hire them, and how to work FDE-style.
• What is an agent harness?An agent harness is everything around the model that makes it an agent: instructions, tools, memory files, the loop and the stop rules. What it is, why agents fail without one, and an example.
• What are AI evals?An eval is a test for an AI system: an input, and a way to grade the output. What evals are, the three kinds of grader, why a small team needs 20 test cases, and an example.
• What are AI skills?An AI skill is a folder with a SKILL.md file: a name, a description and instructions an AI loads when a task needs them. The open standard, who supports it, and how a team shares one.
• What are AI agents?An AI agent is a model that decides its own steps and uses tools until a job is done. How agents differ from chatbots and workflows, where they fit a small team, and an example.
• Prompt vs context engineeringPrompt engineering is writing the instructions; context engineering is choosing everything the AI sees. The difference, when each matters, and how a team keeps good prompts.
• What is a vector database?A vector database stores embeddings so an AI can find text by meaning, not exact words. How it works, when a small team needs one, and what to prepare before choosing.
• Fine-tuning vs RAGRAG gives an AI your current facts at answer time; fine-tuning changes the model with training examples. The difference, what OpenAI now says about fine-tuning, and how to decide.
• AI governance policyAn AI governance policy says which AI tools your team may use, what data never goes in, who checks output, and who owns it. What the frameworks say, and a one-page start.
• LLM evaluation for non-engineersHow to evaluate what an LLM gives you without code: check claims, sources, tone and gaps before you use a draft, and when to build a test set. With an example and a checklist.