AI is modernizing the way we work and moving the industry forward at an unprecedented pace. AI has especially seen accelerated growth in the last 5 years, with the transformer architecture driving major progress to a new era of modern AI. As AI continues to evolve, new terms and concepts are becoming part of everyday business conversations. Understanding what these terms mean and why they matter can help teams keep up with the latest developments in AI. Here are some of the key terms worth knowing.
A transformer is a type of AI model architecture behind many of the biggest advances in AI. It helps models understand context and language by identifying which words and phrases are most important in relation to one another. Transformers were introduced in 2017 by Google researchers in the landmark paper Attention Is All You Need and have since become the foundation of modern AI. Unlike earlier AI models that processed text one word at a time, transformers can consider the relationships between all the words in a sentence at once, allowing them to better understand context, meaning, and intent.
LLMs (Large Language Models) are a type of AI trained on vast amounts of text to understand language, generate content, answer questions, and respond to user prompts. Popular examples include ChatGPT, Claude, Gemini, and other AI assistants. Rather than searching for predefined answers, LLMs identify patterns in the data they were trained on to understand context and predict the most likely next words in a response. Most modern LLMs are built using transformer architecture, which enables them to understand the relationships between words and produce coherent, human-like text.
Generative AI is a type of AI that creates new content based on patterns learned from the data it was trained on. Depending on the model, it can generate text, images, videos, audio, code, and other types of content. LLMs are a subset of generative AI that specialize in understanding and generating human language. In recent years, generative AI has become widely adopted through tools like ChatGPT, Claude, Gemini, Midjourney, and other AI-powered applications.
An AI agent is an AI system that can do more than respond to prompts. It can autonomously plan, make decisions, use tools, and complete tasks on a user's behalf. Rather than simply generating an answer, an AI agent takes action to achieve a specific goal.
For example, if an order you placed is delayed, a traditional chatbot might only notify you of the delay. An AI agent could investigate the issue, check the latest shipping status, contact the carrier, recommend solutions, and keep you updated throughout the process.
AI agents are increasingly being used to automate workflows, improve customer support, and deliver more personalized digital experiences.
In short, chatbots answer questions, while AI agents can take action. See our recent post on AI agents to dive deeper.
MCP (Model Context Protocol) is an open standard that allows AI applications to securely connect with external tools, data sources, and software. Without MCP, an AI model is generally limited to the information provided in a user's prompt. With MCP, it can access resources such as company documents, Google Drive, databases, calendars, websites, and other connected systems to retrieve relevant information and perform tasks. By giving AI access to the tools and data it needs, MCP enables more accurate responses, better context, and the ability to automate multi-step workflows with less manual effort.
RAG (Retrieval-Augmented Generation) is an AI technique that allows a model to search external sources for relevant information before generating a response. Instead of relying only on what it learned during training, RAG can retrieve up-to-date or organization-specific information from documents, websites, knowledge bases, and other connected data sources. This helps AI provide more accurate, relevant, and context-aware responses, especially when working with large amounts of information.
RAG and MCP are complementary technologies. MCP provides the connection to external tools and data sources, while RAG retrieves the most relevant information from those sources so the AI can use it when generating a response.
A context window is the amount of information an AI model can "see" at one time. This includes the prompt, conversation history, and any uploaded files or images. Since this amount is limited, AI can't analyze an entire knowledge base or document library all at once. RAG helps by finding the most relevant information from external sources and adding it to the context, allowing the AI to generate more accurate and informed responses.
Multimodal AI is a type of AI that can understand and work with multiple forms of information, not just text. Depending on the model, it can process images, audio, video, documents, and other types of media alongside written prompts. For example, you can upload an image, ask a question about it, or request changes, and the AI can use both the visual and written information to generate a response.
AI-powered personalization uses AI to analyze a user's behavior, preferences, and interactions to deliver a more relevant and tailored experience. By recognizing patterns over time, AI can predict what content, products, or information is most likely to interest each individual. For example, an e-commerce website can recommend products based on a user's browsing and purchase history, while social media platforms can personalize content by learning which posts, videos, or topics a user engages with most.
AI workflow automation uses AI to automate multi-step business processes, reducing the need for manual work and human intervention. Rather than handling a single task, AI can analyze information, make decisions, use connected tools, and move a process forward from one step to the next. For example, when a customer submits a support request, AI can understand the issue, search internal knowledge bases for the best solution, draft a response, route it to a human for approval if needed, and then send the final reply to the customer. By automating repetitive and time-consuming tasks, AI workflow automation helps organizations improve efficiency, reduce errors, and free employees to focus on higher-value work.
AGI (Artificial General Intelligence) is a form of AI that is capable of understanding, learning, and performing a wide variety of intellectual tasks, rather than being designed for a specific purpose. Unlike today's AI systems, which excel at tasks such as writing, coding, analyzing data, or generating images, an AGI could adapt to entirely new problems without being specifically trained for each one.
For example, if asked to solve a math problem, an AGI could not only explain the solution but also connect it to concepts in physics, economics, engineering, and other fields, applying knowledge across disciplines. In short, while today's AI is designed for specific tasks, AGI represents the idea of AI that can learn, reason, and adapt across many different domains in a way that is more similar to how humans think and solve problems.
This is just a high-level overview of some of the most common AI terms you'll encounter. AI is a rapidly evolving field, with new technologies, concepts, and terminology emerging all the time. Having a solid understanding of these fundamentals makes it easier to follow industry developments and recognize how AI is transforming the way businesses operate.
At Perpetual, we leverage these technologies to build intelligent, user-centered digital products that help our clients solve complex problems, improve experiences, and stay ahead in a rapidly changing landscape.
