Glossary

An agent is a component of the AI system responsible for interpreting user input and executing tasks or actions in response. Agents can handle a variety of tasks, from answering questions to performing specialized actions like sending notifications or analyzing data. Agents work within the broader AI system, often interacting with other components like databases or user interfaces to deliver seamless user experiences.

An AI system designed to assist users in managing tasks, retrieving information, or facilitating conversations. AI Assistants are commonly integrated into workflows to enhance productivity and efficiency by offering consistent, real-time support.

A series of interactions between a user and the AI Assistant, typically involving multiple related tasks. The AI Assistant uses context from previous exchanges to maintain coherence and respond appropriately as the chat progresses. During a chat, the user might request the assistant complete multiple tasks such as planning an itinerary or researching restaurants, all within the same thematic context.

Compound AI refers to an advanced AI system that combines multiple AI models, algorithms, or methodologies to create a more capable system than any single LLM could achieve alone. Like a chemical compound that forms from different elements, Compound AI integrates various AI technologies (such as machine learning, natural language processing, computer vision, and inference-level optimization) to handle complex tasks such as logic and coding, and provide stronger output. This approach allows the system to leverage the strengths of each component while potentially compensating for individual weaknesses across LLMs.

The information retained from previous interactions that the AI Assistant uses to maintain continuity and relevance in ongoing conversations. Context helps the AI tailor its responses to align with past user inputs.

The amount of information within a conversation the AI Assistant can retain at any given time. The size of the context window varies depending on the AI’s design and capabilities. In many AI systems, the context window is measured in tokens, which determine how much of the conversation the AI can reference at once.

A record of the conversation between the user and the AI, which can be referenced for continuity.

Deep research is AI-driven approach to gathering, analyzing and consolidating mass amounts of data across the web, documents, and other sources to deliver a comprehensive report. Deep Research is similar to having an AI research assistant or analyst, which uses an LLM built on AI reasoning to develop and execute a research plan for you.

A continuous back-and-forth exchange between the user and the AI Assistant, typically involving multiple tasks within one or more conversations.

GPT stands for "Generative Pre-Trained Transformer". This term encapsulates the core functionalities and architecture of an AI model . This type of AI model is used for natural language processing and is capable of generating human-like text based on extensive data training.

The medium through which users interact with the AI Assistant. Common interfaces include text-based chat windows, voice command interfaces, mobile applications, and code-based integrations (called APIs).

Input provided in everyday conversational language, as opposed to structured commands. The AI Assistant processes natural language input, allowing users to communicate without special formatting or technical instructions.

The input provided by the user to the AI Assistant, which can take the form of a question, command, statement, or instruction. Prompts guide the AI Assistant's actions by serving as the primary input for generating responses or performing tasks.

A process where the user adjusts or rephrases their prompt to obtain a more accurate or relevant response from the AI Assistant. Refinement helps the AI better understand the user's specific needs, ensuring that the output aligns more closely with the desired outcome.

The output generated by the AI Assistant and returned to the user. Responses are tailored to the specificity or generality of the user's prompt, ensuring relevant and accurate output.

The amount of time it takes for the AI Assistant to generate a response after receiving a prompt.

An action performed by the AI Assistant based on the user's prompt. Tasks consist of a user-provided prompt and the AI's corresponding response. Tasks can involve a single action, such as retrieving information, or a series of actions coordinated by the AI Assistant.

The underlying goal or desired outcome a user aims to achieve when interacting with the AI Assistant. Understanding user intent is essential for the AI to generate relevant responses or perform appropriate tasks.

A non-verbal response provided by the AI Assistant, which includes the display of data, charts, images, or other visual elements. A visual response is often used to complement verbal or text-based responses, helping users better understand complex information.

A method of communication where the user interacts with the AI Assistant using spoken commands, typically through a device’s microphone. Voice interaction relies on speech recognition technology to interpret the user's spoken inputs and respond accordingly.