# Glossary

A comprehensive reference of terms and concepts used in the Twig AI platform.

## A

### Agent

An AI-powered assistant configured with specific data sources, instructions, and behavior patterns to answer questions and provide information.

### Agentic Workflow

A mode where the AI agent can use tools, make function calls, and perform multi-step reasoning to solve complex problems.

### API Key

A credential used to authenticate programmatic access to the Twig AI REST API.

## C

### Cedar Strategy

A RAG strategy that rewrites user queries based on conversation context before retrieval. Balanced speed (\~2-3 seconds) and accuracy.

### Chunking

The process of breaking large documents into smaller segments for more precise retrieval and processing.

### Citation

A reference to the source document used to generate an AI response, providing transparency and enabling fact-checking.

### Context Window

The amount of text (measured in tokens) that can be processed in a single request by a language model.

### Cypress Strategy

An advanced RAG strategy using query expansion, tier-based retrieval, and automatic reranking. Highest accuracy (\~3-4 seconds).

## D

### Data Source

A repository of information that can be ingested into Twig AI, such as documents, websites, databases, or cloud storage.

## E

### Embedding

A numerical representation of text that captures semantic meaning, enabling similarity-based search.

### Evaluation (Evals)

Automated tests that measure AI agent performance across metrics like relevance, accuracy, and citation quality.

## H

### Hallucination

When an AI generates information that isn't supported by the provided context or source documents.

## I

### Inbox

The interface where you review, edit, and improve AI responses to train and optimize your agents.

### Interaction

A single question-answer exchange between a user and an AI agent, stored for analytics and improvement.

## K

### Knowledge Base (KB)

A curated collection of articles, automatically generated from interactions or manually created, that serves as truth data.

## L

### LLM (Large Language Model)

The AI model that generates responses, such as GPT-4, GPT-3.5-turbo, or Claude.

## M

### Memory

Conversation history stored to maintain context across multiple turns of dialogue.

## P

### Pinecone

A vector database service used for storing and searching document embeddings.

### Playground

A testing environment where you can interact with AI agents before deploying them to production.

### Private Data

Data sources that contain organization-specific information, accessible only to authorized users.

### Prompt Engineering

The practice of crafting effective instructions and prompts to guide AI behavior and responses.

## R

### RAG (Retrieval-Augmented Generation)

An AI technique that retrieves relevant information from a knowledge base and uses it to generate accurate, contextual responses.

### Redwood Strategy

The fastest RAG strategy using direct vector search without prompt rewriting (\~1-2 seconds).

### Reranking

A secondary ranking step that reorders retrieved documents using a sophisticated cross-encoder model to improve precision.

## S

### Semantic Search

Search based on meaning rather than exact keywords, using vector embeddings to find conceptually similar content.

### Session

A continuous conversation context, identified by a session ID, that allows the AI to remember previous exchanges.

### SSO (Single Sign-On)

Authentication method that allows users to access Twig AI using their organization's identity provider.

### Strategy Code

An identifier specifying which RAG strategy an agent uses (e.g., `STANDARD_RAG`, `INFER_QUESTION`, `CYPRESS`).

## T

### Temperature

A parameter (0.0-1.0) controlling the randomness of AI responses. Lower values are more deterministic, higher values more creative.

### Tier

A priority level assigned to data sources, used in Cypress strategy for organized retrieval.

### Token

A unit of text processed by language models, roughly equivalent to 4 characters or 0.75 words.

### topK

The number of most relevant documents retrieved from the vector database (typically 5-10 for Redwood/Cedar, 50 for Cypress).

## V

### Vector Database

A specialized database for storing and searching high-dimensional embeddings, enabling semantic search.

### Vector Embedding

See **Embedding**.

## W

### Webhook

An HTTP callback that sends real-time notifications when specific events occur in your Twig AI organization.

***

For more detailed explanations, see [Core Concepts](/getting-started/core-concepts.md).


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://help.twig.so/product/reference/glossary.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.