AI Technology Dictionary

An open communication protocol developed by Google, defining standard ways for AI agents to exchange information. Enables agents from different providers and frameworks to collaborate in a standardized way.

Protocol for communication between an AI agent and the user interface (client / frontend). Defines how results, progress events and dynamically generated UI elements from the agent are presented to the end user.

A computer program capable of autonomously making decisions and taking actions to achieve a specified goal. An AI agent can use tools, call APIs, search knowledge bases and communicate with other agents.

A JSON file published by an A2A-compliant agent, describing its capabilities, supported task types, inputs, outputs and authentication requirements. Enables other agents to automatically discover and call the agent's services.

A managed runtime in Google Vertex AI designed to host AI agents (LangChain, LangGraph). Provides scaling, session state management, logging, monitoring and security — without managing your own infrastructure.

Google's cloud platform offering computing, storage, networking, machine learning and AI services. The ANITA & ANTEK ecosystem runs entirely on GCP, using Vertex AI, Cloud Run and Google Drive among others.

Multimodal large language models (LLMs) developed by Google DeepMind. Available in Flash (fast, cost-effective), Pro and Ultra variants. Supports text, code, images and audio. Used as the reasoning engine in ANITA and ANTEK agents.

A library from the LangChain ecosystem for building AI agents as directed graphs (DAGs). Each graph node is a processing step, and edges define the control flow. Enables agents with loops, branching and complex orchestration logic.

An open-source, self-hosted chat platform and advanced alternative to ChatGPT. Supports multiple AI providers, user management, RAG and plugins. Great for internal enterprise deployments.

A Python library and proxy server providing a unified OpenAI-compatible API for 100+ LLM providers. Enables easy provider switching, load balancing and cost logging without changing application code.

A large language model — a type of neural network trained on vast text datasets, capable of understanding and generating natural language. Examples: GPT-4, Gemini, Claude, Llama. The foundation of modern AI agents.

An extensible, self-hosted web interface for AI models that works completely offline. Supports local models via Ollama and remote APIs. Offers custom agent creation, data pipelines, RAG on your own documents and user management.

A technique that enriches an AI model's prompt with document fragments retrieved from a knowledge base (vector store). Allows the model to answer based on current, company-specific data without retraining.

An HTML5 web standard enabling a server to push events to a client over a persistent HTTP connection. Used in the A2A protocol to stream agent task results in real time.

A TypeScript/JavaScript library for building AI user interfaces in Next.js and other frameworks. Provides React hooks, response streaming, agent tools, structured data generation and integrations with many LLM providers.

Google Cloud's managed MLOps platform for training, deploying and monitoring AI models. Includes Agent Engine, Model Garden (model catalog), Vector Search and RAG tools.