AI Glossary
& Tool Directory

Adobe Firefly is Adobe's generative AI, trained on licensed content and thus commercially safe to use. Integrated into Photoshop, Illustrator, and Express, AI steps run directly inside the existing creative workflow.

Agentic Marketing describes the use of autonomous AI agents in marketing workflows. Instead of individual tools, specialized agents handle complete task chains — from market analysis to content creation to performance optimization. The agents work in a loop: Plan → Create → Execute → Learn.

Which AI video generator fits which job? Veo 3, Sora 2, Kling, Luma Ray2, Runway, Hailuo, Seedance and Pika honestly compared: strengths, audio, free tiers, pricing.

APIs enable developers to programmatically integrate AI models into their own applications. Through APIs from OpenAI, Anthropic, or Google, language models, image generation, and more can be directly integrated.

Canva has integrated extensive AI features: Magic Design creates layouts automatically, Magic Write generates text, and Magic Edit enables AI-powered image editing — all within the familiar Canva interface.

CapCut is the video editor behind TikTok with powerful AI features. Auto-captions, background removal, AI effects, and templates make creating social media videos extremely fast.

Chain-of-Thought is a prompting technique where the AI model is asked to explicitly lay out its reasoning steps. This improves quality for complex tasks like math, logic, and multi-step analysis.

ChatGPT is an AI chatbot by OpenAI based on GPT language models. It can write text, answer questions, generate code, and create creative content — from blog posts to marketing concepts.

What Claude can do in 2026: the current model family (Opus, Sonnet, Haiku), pricing, context window, Claude Code and how it compares to ChatGPT and Gemini.

ComfyUI is a visual, node-based editor for Stable Diffusion workflows. Its modular architecture enables building complex image pipelines — from simple generation to elaborate multi-model setups.

Computer vision is the field of AI that enables machines to understand and interpret visual information. From object detection to facial recognition to medical image analysis — the applications are diverse.

Copy.ai has evolved from a pure copywriting tool into a go-to-market platform. It automates content production from social media posts to emails to complete sales sequences.

Cursor is a VS Code-based code editor with deep AI integration. It offers context-aware autocompletion, chat across the entire codebase, and can generate whole functions based on instructions.

DALL-E is OpenAI's AI image generator, directly integrated into ChatGPT. It creates images from text descriptions and is especially strong at precise implementations of specific requirements and infographics.

DeepL is an AI translation service known for its exceptionally natural translations. Especially for European languages, it significantly outperforms Google Translate and also offers DeepL Write as an AI writing assistant.

Descript revolutionizes media editing: videos and podcasts are edited like a text document. Deleting words means cutting scenes, and the AI can even fill in missing words in your own voice.

Diffusion models generate images by gradually removing noise from a random signal. This process produces highly detailed images and is the foundation for Midjourney, DALL-E, and Stable Diffusion.

ElevenLabs is the leading platform for AI-generated speech and voice cloning. The voices sound strikingly natural and support dozens of languages with emotional nuance.

Embeddings are numerical representations of text, images, or other data as vectors in multi-dimensional space. Similar content is positioned close together, enabling semantic search and comparisons.

Few-shot learning enables AI models to solve tasks after receiving just a few examples in the prompt. Instead of extensive training, two to five examples are enough for the model to recognize the desired pattern.

Fine-tuning is the process of further training a pre-trained AI model with custom data. This allows the model to better handle specialized tasks without needing to be trained from scratch.

Foundation models are large, pre-trained AI models that serve as the basis for diverse applications. GPT-4, Claude, and Llama are foundation models that are adapted for specific tasks through fine-tuning and prompting.

Gamma is an AI tool that automatically generates presentations, documents, and websites from text. Instead of manually designing slides, you describe the content — Gamma creates a fully designed, interactive deck with images, diagrams, and animations.

Gemini is Google's AI assistant that can natively process text, image, audio, and video. Deeply integrated into the Google ecosystem, it offers direct access to Gmail, Docs, and Google Search.

GitHub Copilot is an AI-powered coding assistant that delivers real-time code suggestions directly in the editor. Trained on billions of lines of public code, it understands context and patterns in virtually any programming language.

Veo 3 is Google's AI video generator and the first major model to create picture and sound together: dialogue, sound effects and ambience come straight out of the model. Available through Gemini, Flow and the API.

GPT stands for Generative Pre-trained Transformer and is OpenAI's model series powering ChatGPT. The architecture combines pre-trained language understanding with the ability to generate coherent and contextual text.

Grammarly has enhanced its spelling checker with generative AI. The assistant can rewrite text, adjust tone, create summaries, and generate complete text suggestions.

Grok is an AI chatbot by xAI, Elon Musk's AI company. Integrated into the X platform (formerly Twitter), Grok features real-time access to social media data, a humorous communication style, and minimal content filters.

Hailuo AI is the video generator by Chinese AI lab MiniMax. Known for extreme motion physics (acrobatics, stunts, animals), bracket-based camera commands in Director mode and a generous free tier.

Hallucinations occur when AI models generate plausible-sounding but factually incorrect information. The model doesn't have real knowledge but generates statistically probable text — which can lead to convincing misinformation.

HeyGen creates professional videos with AI avatars that speak and gesture naturally. Ideal for training videos, product demos, and personalized video messages without a camera or studio.

Hugging Face is the central platform for open-source AI models. With over 500,000 models, datasets, and Spaces, it's the GitHub of the AI world — from NLP to computer vision to audio.

Ideogram is an AI image generator that excels at precise text rendering in images. While other AI models often fail at text in images, Ideogram can generate logos, posters, and designs with readable, correctly displayed text.

Inference refers to the process of applying a trained AI model to new inputs and generating results. Inference speed and cost are crucial factors for productive AI deployment.

Jasper is an AI platform specifically for marketing teams. With Brand Voice, campaign templates, and team collaboration features, it generates brand-compliant content at scale.

AI agents are systems that can independently plan, execute tasks, and use tools and APIs. They go beyond simple chatbots by autonomously running multi-step workflows.

AI image generation means creating images from text descriptions (prompts) using artificial intelligence. Models like DALL-E, Midjourney, Stable Diffusion, and Flux transform natural language into photorealistic images, illustrations, or designs — without manual graphic work.

Kling is an AI video generator by Kuaishou that creates impressively realistic videos with complex camera movements and consistent characters. It excels at human depiction and has established itself as a leading AI video tool globally.

The context window defines how many tokens a language model can process simultaneously. It determines how long conversations or documents a model can comprehend — from 4,000 to over 1 million tokens.

LangChain is an open-source framework for building applications with language models. It provides abstractions for chains, agents, RAG pipelines, and tool usage, and has established itself as the standard framework for LLM apps.

Large Language Models are AI models trained on massive amounts of text that can understand and generate human-like language. GPT-4, Claude, and Gemini are prominent examples of this technology.

LoRA is an efficient fine-tuning method that trains only small adaptation layers of a model instead of the entire model. This makes specialized training possible even with limited hardware and less data.

What Luma Dream Machine can do in 2026: the Ray2 model, what the free tier includes, pricing, API access, and how Luma compares to Kling, Runway and Veo 3.

Make (formerly Integromat) is a no-code platform for complex automations. The visual builder connects APIs, databases, and AI services into powerful workflows.

A Marketing Operating System (Marketing OS) is an integrated platform of AI agents, workflows, and data pipelines that connects all marketing processes of an organization. Unlike individual SaaS tools, a Marketing OS provides a unified architecture where agents collaborate autonomously — from planning to analysis.

Meta AI is the AI assistant by Meta, integrated into Facebook, Instagram, WhatsApp, and Messenger. Based on the Llama models, Meta AI offers text generation, image generation, and web search directly within the platforms billions use daily.

Midjourney is an AI image generator known for its exceptionally aesthetic and artistic results. It operates through Discord, and with the right prompts, it produces images in virtually any style.

Mistral AI is a French AI company developing powerful open-source language models. Mistral models offer excellent performance-to-efficiency ratios and are an important European alternative to OpenAI and Google.

Multimodal AI processes different data types simultaneously — text, image, audio, and video. Models like GPT-4o and Gemini can analyze an image and discuss it, enabling entirely new applications.

n8n is an open-source automation platform that can be self-hosted. It offers the flexibility of code with the accessibility of a visual editor and can be extended with custom nodes.

Neural networks are computational models inspired by biological neurons, consisting of layers of connected nodes. They learn from data by adjusting weights and form the foundation for deep learning and all modern AI systems.

Natural Language Processing encompasses all AI techniques for understanding and generating human language. From sentiment analysis to named entity recognition to machine translation — NLP is the foundation of all language models.

NotebookLM by Google analyzes uploaded documents, allowing you to ask targeted questions and generate summaries. It even creates audio summaries in a podcast format.

OpenAI is the AI research company behind ChatGPT, GPT-4, DALL-E, and the OpenAI API. Founded in 2015, OpenAI has transformed how businesses worldwide work with artificial intelligence through its Large Language Models.

Opus Clip automatically transforms long videos into viral short-form clips. The AI identifies the best moments, adds subtitles, and optimizes formatting for TikTok, Reels, and Shorts.

Perplexity is an AI-powered search engine that delivers answers with verified citations. It combines the naturalness of a chatbot with the reliability of a research engine.

Pika is an AI video generator distinguished by especially creative and stylized results. With features like Scene Expansion and Lip-Sync, it offers unique creative possibilities.

Prompt engineering is the technique of formulating instructions to AI models to achieve optimal results. It encompasses strategies like role assignments, examples, formatting guidelines, and step-by-step instructions.

RAG combines large language models with external data sources. Instead of relying solely on training data, the model can retrieve current and specific information from documents, databases, or APIs.

Replicate makes it easy to use open-source AI models through a unified API. Thousands of models run in the cloud without needing to operate your own GPU infrastructure.

RLHF is a training method where AI models are improved through human feedback. Human evaluators rank different model responses, and the model learns to prefer helpful and safe answers.

Runway is an AI platform for video generation and editing. With Gen-3 Alpha, it creates impressive videos from text or images and offers additional tools like greenscreen removal and motion tracking.

Seedance is ByteDance's AI video model. Its unique feature: native multi-shot generation. One prompt returns an edited sequence of several shots with consistent characters.

Semrush has deeply integrated AI into its marketing suite. From AI-generated SEO recommendations to content optimization to competitive analysis, the platform uses machine learning for data-driven marketing.

Sora is OpenAI's AI model for video generation that creates realistic videos from text descriptions. It understands physical relationships and can generate complex scenes with consistent motion.

Stable Diffusion is an open-source image generation model that can run locally or in the cloud. Its open architecture has spawned a massive ecosystem of models, LoRAs, and extensions.

Suno generates complete songs with vocals, instruments, and production from text descriptions. Results range from pop to rock to hip-hop and sound surprisingly professional.

Surfer SEO analyzes top-ranking results for keywords and provides data-based recommendations for content structure, word count, and keyword density. The integrated AI writer generates SEO-optimized articles.

Synthesia is an enterprise platform for AI-generated videos with photorealistic avatars. Especially popular for corporate communication, onboarding, and training in over 140 languages.

Temperature is a parameter that controls the randomness of AI output. Low values (0.1) produce predictable, consistent responses; high values (1.0+) lead to more creative but less predictable results.

Tokens are the smallest text units that a language model processes — often word parts or characters. They determine the cost, context length, and processing speed of AI requests.

Training data is the data used to train AI models. The quality, volume, and diversity of training data significantly determine the performance and potential biases of the finished model.

The Transformer architecture is the foundation of virtually all modern AI language models. Introduced by Google in 2017, it uses the attention mechanism to understand relationships in data — regardless of sequence order.

Udio is an AI music generator that produces songs in studio quality. Especially strong with complex arrangements and nuanced genres, it offers fine control over style, mood, and structure.

Vector databases store and search high-dimensional vectors efficiently. They are the backbone of RAG systems and enable lightning-fast semantic search across millions of documents.

Writesonic is an AI content platform that generates everything from blog articles to landing pages to product descriptions. With its integrated Chatsonic chatbot, it also offers a ChatGPT competitor.

Zapier connects over 6,000 apps and has leveled up with AI features. The AI Chatbot Builder and Natural Language Automation let you create workflows via voice commands.

Zero-shot learning describes the ability of AI models to solve tasks without having seen examples beforehand. The model uses its general language understanding to handle new task types.