Oracle Agentic AI Foundations key terms

A software entity that perceives its environment, makes decisions, and takes actions to achieve goals, often autonomously and with minimal human intervention.

AI systems designed to act with agency: setting subgoals, planning, using tools, and adapting behavior to accomplish complex tasks over time.

An agent that operates independently, making decisions and executing actions without continuous human oversight, within defined boundaries.

A network of multiple agents that interact, coordinate, or compete to solve problems or perform tasks that are beyond the capability of a single agent.

The coordination and management of multiple agents, including task allocation, communication, and conflict resolution, to achieve a collective goal.

The exchange of messages or signals between agents to share information, coordinate actions, and negotiate, often using standardized protocols.

The structural design of an agent, including its components such as perception, reasoning, memory, and action modules, and how they interact.

A blueprint for intelligent agents that specifies memory, learning, reasoning, and decision-making mechanisms inspired by human cognition.

Oracle's integrated cloud service for building, deploying, and managing AI agents that connect to enterprise data and applications.

Oracle's cloud computing platform offering compute, storage, networking, and AI services used to host and run AI agents.

A managed Oracle Cloud service that provides access to large language models for text generation, summarization, and other generative tasks via API.

A low-code development environment within Oracle AI Agent Platform for visually designing, testing, and deploying AI agents.

A deep learning model trained on vast text data to understand and generate human-like language, often used as the reasoning engine in agents.

The practice of designing and refining input prompts to guide an LLM's output toward desired responses, improving accuracy and relevance.

A technique that encourages an LLM to explain its reasoning step by step, often improving performance on complex tasks.

An external function, API, or service that an agent can invoke to extend its capabilities, such as web search, calculators, or database queries.

A mechanism where an LLM outputs structured requests to invoke predefined functions, enabling integration with external systems and tools.

The process of connecting an agent to external applications or data sources via application programming interfaces to enable actions and data retrieval.

A technique that enhances LLM responses by retrieving relevant information from external knowledge sources and incorporating it into the generation process.

A database optimized for storing and querying high-dimensional vectors, enabling similarity search for unstructured data like text and images.

A numerical representation of data (e.g., text) as a vector in a continuous space, capturing semantic meaning for tasks like similarity search.

A structured repository of information, often in the form of documents or databases, that an agent can query to inform its responses.

A search technique that uses meaning and context rather than exact keywords to find relevant information, often powered by embeddings.

Safety mechanisms that constrain an AI agent's behavior to prevent harmful, biased, or inappropriate outputs and actions.

An approach to AI development and deployment that emphasizes fairness, transparency, accountability, privacy, and safety.

The policies, processes, and controls that ensure AI systems are developed and used in compliance with regulations and ethical standards.

A design pattern where humans review, approve, or correct AI agent decisions at critical points to ensure oversight and accuracy.

Systematic and unfair prejudice in AI outputs, often stemming from biased training data or flawed algorithms, leading to discriminatory outcomes.

The principle that AI systems should treat all individuals and groups equitably, without discrimination or favoritism.

The ability to understand and articulate how an AI agent arrived at a decision or output, fostering trust and accountability.

The principle that AI systems should be open about their capabilities, limitations, and decision-making processes to relevant stakeholders.

The stages an AI agent goes through from conception to retirement: design, development, testing, deployment, monitoring, and decommissioning.

The continuous observation of an AI agent's performance, behavior, and resource usage in production to detect issues and ensure optimal operation.

The practice of managing multiple versions of AI models and their associated artifacts to track changes and enable rollback.

The process of making a trained AI model available for use in production, often involving packaging, scaling, and integration with applications.

The recording of events, interactions, and errors during an AI agent's operation for debugging, auditing, and performance analysis.

A chronological record of an AI agent's activities and decisions, used for compliance, security, and forensic analysis.