A repository where an agent stores past interactions and outcomes to inform future decisions.

Checking outputs for correctness before accepting them, often to catch errors or reward hacking.

Learned capabilities that enable AI agents to perform specialized tasks with improved effectiveness.

Compressing or summarizing an LLM's past context into a smaller representation to extend memory and improve reasoning within limited token budgets.

An AI-powered software system that autonomously writes, modifies, or extends source code based on high-level instructions.

A system where multiple AI agents collaborate, each handling specialized roles, communicating and coordinating to complete complex tasks.

A system that structures, verifies, monitors, or controls how an AI agent interacts with external tools, environments, or execution runs.

Training approach that lets LLM-based agents learn multi-step tool use and planning via reinforcement learning, often with dense or per-tool rewards.

Measuring AI agents' performance in complex tasks, highlighting failures in planning, monitoring, and real

A model that handles tasks combining visual and textual information, such as editing images from conversation or reasoning across images and text.

A method that retrieves relevant external information to ground and improve the output of a generative model.

Risks from AI agents including timing errors, shutdown avoidance, collusion, overclaiming, and unsafe execution, mitigated by constrained decoding, safety harnesses, and swarm governance.

Identifying which component or step in a system caused an observed failure, often using structured trace representations.

The collaborative operation of multiple AI agents to solve tasks more effectively or with greater capability.

Iteratively adapting an agent's environment, prompts, or evaluation setup to improve task performance and generalization.

Determining which actions or steps in a sequence contributed to a final outcome, often for reinforcement learning.

A standard set of tasks or data used to evaluate and compare the performance of AI systems.

The capability of an agent to invoke external tools sequentially over multiple conversational turns to complete complex tasks.

Tracking state with explicit data types, separated from

Summarizing or structuring past information to reduce token usage while preserving task accuracy.

Having access to relevant evidence does not prevent an LLM agent from making incorrect or unsafe decisions.

Spending more computation during inference, often by running multiple agent iterations or expanding context, to improve model output quality.

A granular assessment of a system's intermediate steps or subgoals, rather than just its final output, to diagnose specific failures.

Autonomous agents automatically generate, control, or extend simulations to improve accuracy, interactivity, or performance.

An AI system repeatedly uses its own outputs to enhance its capabilities, creating a feedback loop for autonomous improvement.

Directing each query or subtask to the most suitable model to optimize cost, accuracy, or quality.

A multi-step AI system that autonomously gathers, evaluates, and synthesizes evidence from diverse sources to produce comprehensive reports.

Feedback that evolves alongside the policy or components it guides to improve performance.

A reinforcement learning method that optimizes policy by computing advantages based on group-level reward statistics.

A learned internal, compressed representation of environment dynamics, enabling future-aware planning, generation, and reasoning directly in latent space.

An agent that improves by reusing its own experience, memory, or self-generated feedback, with success depending on verification and design.

A paradigm where actions are expressed as executable code, such as method calls or program synthesis, rather than direct manipulation.

Reinforcement learning that uses rewards from objectively checkable signals like correct answers or verification feedback.

A robot policy model that maps visual observations and language instructions to actions, trained on demonstrations and often used for manipulation tasks.

Mapping natural language commands or descriptions to specific visual elements of a user interface.

A predictive environment model that agents use to simulate outcomes, enabling planning, reasoning, and long-horizon task execution.

AI systems that interact with the physical world through sensors and actuators in a body or robot.

Multiple autonomous agents coordinate and combine their specialized capabilities to solve complex tasks beyond individual ability.

A neural architecture with many specialized sub-networks, only a few activated per input via a router, balancing capacity and efficiency.

A step-by-step reasoning process that produces auditable trajectories for verification and accuracy.

A multi-step process that generates or verifies spatial arrangements of objects against explicit constraints and geometry.

A post-training method that adjusts model behavior using comparisons or self-generated preferences to improve alignment.

The ability of a system to preserve and recall information over time or across contexts, often balancing compression against fidelity.

Multi-agent systems collaborate autonomously to produce long-form videos maintaining narrative and visual coherence.

A directed knowledge structure where skills are linked, enabling targeted updates and efficient retrieval for AI agents.

The process of developing, refining, or optimizing reusable capabilities in LLM agents through learning, adaptation, or experience accumulation.

Robot manipulation tasks that require completing a sequence of many interdependent sub-tasks over an extended time horizon.

A structured web of entities and relations, used to store domain facts and support reliable reasoning, retrieval, and model alignment.

A learned model generates plans directly for new instances, replacing expensive per-instance search with a fast, generalizable forward pass.

Same inputs give different outputs across runs, confounding testing and evaluation.

External supports like training curricula, tool selection, and success verifiers enable agents to handle long-horizon or out-of-distribution tasks.

In policy optimization, a method that calculates the relative value of actions using group statistics or baselines to guide learning.

Reasoning precisely about object positions, orientations, and spatial relations across planar, depth, and temporal dimensions for robotics and navigation.

Organizing a document corpus into pre-built, reusable navigation views that reduce online retrieval cost and improve accuracy.

A single end-to-end model processes and generates multiple data types (e.g., images, text) with equal competence.

Evaluation that assesses model performance by considering the real-world impact or risk of errors, not just standard accuracy.

Software whose behavior is driven by an embedded AI agent, enabling dynamic adaptation, autonomous task execution, and contract-based integration.

A system for building simulators by combining and extending reusable components, often via automated agents.