Glossary
Short definitions of the terms you see in the app. For how they fit together, see Key concepts.
General
Section titled “General”| Term | Definition |
|---|---|
| Annota AI | The company and the app that contains Annota Studio and Annota Brain. |
| Annota Studio | The product for datasets, image annotation, Documents & OCR, dataset versions, and models. |
| Annota Brain | The product that answers questions from your document knowledge bases, with sources. |
| Organization | The workspace that holds people, data, plan, and credits. You can belong to several. |
| Owner | The member who created the organization, or received ownership. Can delete it and transfer ownership. |
| Admin | A member who manages members, groups, billing, and settings, and moderates every dataset, model, and skill. |
| Member | A person in the organization who works on their own items and on those shared with them. |
| Group | A named set of members, used to share items with several people at once. |
| Invitation | An email that lets someone join an organization. It expires after 7 days. |
| Sharing | Giving members, groups, or the organization Can view or Can edit access to an item. |
| Private | Visible only to its owner. Datasets, knowledge bases, models, and skills start private. |
| Resources | The admin page that lists every dataset, model, and skill in the organization. |
| Plan | The subscription of an organization. It sets monthly credits and capacity. |
| Credits | The unit that paid actions use. One balance per organization, shared by both products. |
| Monthly credits | Credits included in the plan. Used first and do not roll over. |
| Purchased credits | Credits you buy. They never expire. |
| Weekly usage | A per-member Brain allowance, used before the organization credits. |
| MCP server | The connection that lets AI assistants work with your Annota AI data through the Model Context Protocol. |
Annota Studio
Section titled “Annota Studio”| Term | Definition |
|---|---|
| Dataset | A collection of files of one type: images or documents. |
| Image dataset | A dataset of images that you annotate with bounding boxes and polygons. |
| Document dataset | A dataset of PDFs, scans, Office and text files, and recordings, converted to text. |
| Asset | One file in a dataset: an image or a document. |
| Live dataset | The current, editable state of a dataset, as opposed to a saved version. |
| Label | A category you annotate, such as person. Each label has a name and a color and belongs to a dataset. |
| Annotation | One shape on an image, linked to a label. |
| Bounding Box | A rectangular annotation around an object. |
| Polygon | An annotation that follows the outline of an object point by point. |
| Magic Select | Outlines the object you click on, using AI. Uses credits. |
| Magic Annotate | Finds and annotates every instance of the selected labels in one image or in the whole dataset. Uses credits. |
| Confidence threshold | The minimum confidence a Magic Annotate prediction needs to be kept. |
| Review status | The state of an asset: To annotate, Pending review, Approved, or Rejected. |
| OCR | Optical character recognition: extracting text from PDFs and scanned pages. |
| Text extraction | Turning a document into text, through OCR, transcription, or conversion. |
| Transcription | Turning an audio recording into text. |
| Dataset version | An immutable copy of a dataset’s ready assets and annotations, used for reproducible exports and training. |
| Export | A downloadable archive of a dataset version or of the live dataset, in a format such as COCO or Markdown ZIP. |
| Model | A container for the trained versions of one task, such as object detection. In private beta. |
| Training run | One training job that creates a new model version from a dataset version. |
| Model version | The result of a successful training run, which you can download. |
Annota Brain
Section titled “Annota Brain”| Term | Definition |
|---|---|
| Knowledge base | A searchable index of one document dataset. Brain answers from the knowledge bases you select. |
| Indexing | Preparing a document so Brain can search it. Uses credits for each page of a new or changed document. |
| Sync | How a knowledge base follows its dataset: new and changed documents are indexed in the background. |
| Approved documents only | A knowledge base setting that indexes only documents with the Approved review status. |
| Conversation (chat) | A series of questions and Brain answers. Private to you. |
| Mode | How Brain answers: Fast answers at once, Think reasons further but is slower. |
| Source | A document page that Brain used for an answer. |
| Citation | A numbered reference in an answer that opens the source document at the cited page. |
| To check | A mark on parts of an answer that Brain could not check against the sources. |
| Summary | A short overview and key facts that Brain generates for each indexed document. |
| Skill | Instructions that Brain follows when a request matches the skill’s description. |
Next steps
Section titled “Next steps”© 2026 Annota S.r.l.