AI makes more sense when you can see it.
Follow a document as it becomes useful information. See how an agent chooses its next step. Trace an answer back to its source. Ten important ideas, explained with diagrams and everyday examples.
Find the idea you want to understand.
Plain-language explanations. Diagrams that do the explaining.
Showing 10 of 10 concepts
Language models
A language model produces text from patterns it learned during training and the information supplied with a request. It can explain, summarise and draft; it can also be wrong.
Explore the conceptSmall language models
A small language model has fewer internal parameters than a larger model. It can be useful for focused tasks with a smaller computing footprint, but it still needs testing on the actual work.
Explore the conceptAI agents
An AI agent uses a model to choose steps and use permitted tools toward a goal. It can look up information and prepare work, with limits on what it may do and when a person must approve.
Explore the conceptAgent orchestration
Agent orchestration coordinates several AI steps or agents: who handles each part, what information moves between them and how the results become one usable output.
Explore the conceptDocument intelligence
Document intelligence turns information in files into something a team or system can use. It can recognise text, understand layout, extract named fields and flag values that need checking.
Explore the conceptOptical character recognition
Optical character recognition, or OCR, turns text in an image or scanned page into machine-readable characters. It gives software something to search and process, but the characters it produces can still be wrong.
Explore the conceptGrounded answers
A grounded answer is based on relevant source information supplied to the model. In a company assistant, that often means finding permitted documents first, then answering with references the reader can check.
Explore the conceptKnowledge graphs
A knowledge graph represents things and the relationships between them. Instead of keeping a supplier, contract and department as isolated records, it can show how they connect.
Explore the conceptMultimodal AI
Multimodal AI works with more than one kind of information, such as text and images. The supported combinations depend on the model; the term does not mean every model can handle every file type.
Explore the conceptPrivate AI
Private AI is an arrangement in which access to AI systems and the data they use is controlled for an organisation. Where the application, files and model run—and what leaves that boundary—determines what “private” means in practice.
Explore the concept