Use this page when a word or label in DataMind OS is unfamiliar. Terms are listed alphabetically, each with a one-sentence definition and a link to the page that explains it.
Admin: the area where administrators manage users, roles and platform settings. The sidebar labels it Admin, and the Dashboard card says Administration. See Administration.
Alert: a SQL query that runs on a schedule and emails people when its condition is met. Create one in Observability > Alerts with New alert. See Create an alert.
Analytical Mind: a Mind that answers questions over the tables you link to it on its Sources tab. See Minds.
API token: a personal token that lets a script or tool call DataMind OS as you, within the scopes you choose. It is shown once, when you create it. See API tokens.
Artifact: a saved output of a chat, such as a query, a report or a slide deck. Find them under Minds > Artifacts, or press Ctrl+Shift+A. See Artifacts.
Autosave: how often the Slides editor saves a version on its own. The options are Never, On every change and timed intervals. See Saving and version history.
Batch size: how many nodes of one workflow run may work at the same time. Set it in the Batch Size field of a workflow. See Create and edit a workflow.
BI table: a table in a semantic model that is bound to one Lake or Warehouse table. See Semantic models and BI tables.
Connection Settings: the Integrations panel that shows the Host, Port, Email and JDBC Connection String a client tool needs. See Connection settings.
Console: one SQL tab in the Editor. New Console (Ctrl+Shift+Q) opens another, and Find & reopen a closed console brings back one you closed. See Consoles, saving and history.
Core table: the BI table at the hub of a semantic model's star schema. Every relationship must involve it. See Relationships and the core table.
Custom visual: a DataPrism visual written in HTML, CSS and JavaScript, or built from a template. See Custom visuals.
Dashboard: your home page after sign-in, with the Chat, Playground and Flow views. See The Dashboard. The Lake also has a Dashboard tab, which shows your datasources as cards. See The Lake screen.
Data mode: whether a semantic model's queries read Clear Values or Tokenized PII. It is fixed when the model is created. See Concepts and glossary.
Data source: an external system connected to the Lake, such as a database. Some labels write it as one word, for example Add datasource. See Supported connectors.
DataMind FM: the live music station on Universe. Its mini player keeps it playing as you move between pages. See Listen to DataMind FM.
DataPrism: the reporting area, where you build reports and explore semantic models. See DataPrism.
Deck: a Slides presentation: a list of slides plus settings for the whole deck, such as size, theme and fonts. The UI says "presentation" in some dialogs and "deck" in many labels. See Slides.
Deploy: create a Lake or Warehouse table physically so that it can be loaded. A new Lake table shows the banner "This table is ready to deploy. Deploy to access settings and lake preview." See Configure a table.
Destination: an external system that Writeback sends data to, such as a database, Google BigQuery or RabbitMQ. Add one with Add destination. See Add and manage destinations.
Dictionary: a Warehouse table of lookup values, filled by SQL queries on Lake tables and refreshed on a schedule. Create one with New dictionary. See Dictionaries.
Dimension: a Warehouse table of descriptive data, such as customers, products or dates, that facts join to. Create one with New dimension. See Create a dimension or fact table.
Documentation: the area where you read the DataMind documentation inside DataMind OS; every signed-in user sees it in the sidebar. See Documentation.
Downstream: in Lineage, the objects that depend on the selected object. The opposite is Upstream. See Explore and trace lineage.
Editor: the SQL area, where you query and explore your data. See Editor.
Fact: a Warehouse table of measurable events, such as sales or orders. Create one with New fact. See Create a dimension or fact table.
Failure alert: a notification set on a workflow, so that you hear when one of its runs fails. See Failure alerts.
Field parameter: a list of fields or measures that a slicer lets report readers switch between. See Field parameters.
Folder: in the Lake tree, a group that holds datasources so a long list stays tidy. Create one with Add Folder. See The Lake screen.
Impersonation: viewing DataMind OS as another user sees it, with Impersonate user in the account menu. A banner shows Viewing as and the person's name, with an Exit button. See Impersonate a user.
Incident: in System Overview, a period when a service's metric rises well above its normal level. It is marked Critical or Warning. See System Overview.
Job run: one run of a workflow. The runs table shows its Status, Start Date, End Date and Duration. See Job runs.
Kartlos: DataMind's AI, offered in model pickers as Kartlos Standard and Kartlos Premium. See Choose models and AI behaviour.
Lake: where raw data from your sources lands before it is shaped. See The Lake.
Lake Builder: a Mind that imports source tables into the Lake and manages their connectors, syncs and load settings for you. See Warehouse Builder and Lake Builder.
Lake table: a source table imported into the Lake, with its own sync type, columns and load history. See Configure a table.
Lineage: the map that traces every table from source to finish. Its page is titled Data Lineage, and it shows only what you may read. See Lineage.
Logical model: a Warehouse grouping object, added from the Warehouse add menu. See Semantic sources, source systems and logical models.
Main group: in DataPrism row-level security, the members who bypass the rules and see every row. See Row-level security and View as.
Measure: a named SQL aggregation, such as a sum of a column, that every report on the model can reuse. See Measures.
Mind: an AI agent you chat with about your data. See Minds.
Model AI: the DataPrism AI panel that reads and changes a semantic model, with your approval. See Model AI.
Node: one step of a workflow run, such as loading one table or running one transformation. Double-click a run to see its nodes and their statuses. See Nodes in a run.
Observability: the area for platform health and performance, with the tabs System Overview, Query Management, ClickHouse, Logs, Mind Analytics, Live Metrics and Alerts. See Observability.
Owner Role: the role a Mind belongs to, set when you create it. It decides which tables the Mind can be given. See Create a Mind.
Pass Data Type: for each table linked to a Mind, how much the AI receives: Structure Only, Sample Data or Full Data. See Link data to a Mind.
Permission: one right to do something, such as view Lake objects or add a Mind. Roles carry permissions, and the product hides what you have no permission for. Your profile lists yours. See How access works.
Permission group: the roles listed on a Warehouse table or column that control who can read it. Admin access is always applied. See Warehouse permissions and row-level security.
Permission tier: how permissions are grouped in a role: High-Level presets that bundle related permissions, Standard task-focused sets, or Granular single permissions. See How access works.
PII: personal data, such as names, email addresses or card numbers. An administrator chooses what happens when it is found in what users send to AI: Block User Traffic, Warn User or Tokenize. Lake and Warehouse columns carry a PII Domain setting. See PII detection.
Popup: a floating window that opens a table or an editor over the page you are on; several can be open at once. Open Popups (Alt+Q) finds one. See Popup windows and Open Popups.
Primary key: the column or columns that identify a row. A Lake table needs one before it can be deployed. See Configure a table.
Public link: a DataPrism link that anyone can open without signing in. You create it under Public links in the share dialog. See Share a report.
Recovery code: a single-use backup code for two-factor sign-in. Each set holds 10 codes. See Two-factor authentication.
Relationship Diagram: a picture of how your tables connect. Open it from the Relationship Diagram tab in the Lake header. See Relationship diagram.
Restart Count: a workflow setting that sets how many times a run that failed on a temporary error is restarted automatically. See Monitor and fix a run.
RLS (row-level security): rules that limit which rows a user sees in a table. In Lake and Warehouse table popups it is the step called RLS. See Warehouse permissions and row-level security.
Role: a named set of permissions and table access that an administrator gives to users. A new installation starts with the built-in admin role only, and administrators create the others. See Roles and permissions.
Scheduled task: a Mind prompt that runs on a schedule, by hand, or when a watched workflow fails or finishes. Open them under Minds > Scheduled Tasks, or press Ctrl+Shift+L. See Scheduled Tasks.
Semantic model: the DataPrism layer that reports are built on. It holds BI tables, relationships, measures, field parameters, row-level security and documentation. See Semantic models and BI tables.
Semantic source: a Warehouse group of source systems, so that every table can be traced back to where its data comes from. See Semantic sources, source systems and logical models.
Slicer: a DataPrism visual that picks values of one field and uses them to filter the other visuals. See Filters, slicers and interactions.
Source system: an upstream data origin linked to one semantic source. See Semantic sources, source systems and logical models.
SSO (single sign-on): signing in through your company's identity provider with Sign in with SSO. See Sign in and sign out.
Staging table: in a Lake table's settings, another name for a Lake table, for example in "Optional notes about this staging table…". In Observability > ClickHouse, Empty staging tables means the platform's working tables for loads and transformations instead. See Configure a table.
Static table: a Warehouse table of reference data that you type or upload yourself, such as country codes. It has no live source. Create one with New static. See Static tables.
Subscription: a schedule that emails a deck as a PowerPoint file, or a report as PowerPoint, PDF or Excel. See Email subscriptions for decks and Email subscriptions for reports.
Sync Log: the load history of a Lake table. See Run, stop and troubleshoot loads.
Sync Schema: a datasource action that re-reads tables and columns from the source and finds new ones. See Run, stop and troubleshoot loads.
Sync type: how a Lake table loads. The options include Full Reload (Truncate Insert), Incremental (Upsert), Full Snapshot and Onetime. See Choose a sync type.
Synchronize: run one load of a Lake table now, with the Synchronize button in the table's settings. See Run, stop and troubleshoot loads.
Training card: a note, written in Markdown, that a Mind reads as instructions or knowledge. You manage them in the Training Hub tab. See Train a Mind with training cards.
Transformation: the query that fills a Warehouse table. Check it with Test Run, confirm it with Finalize, then run it with Execute. See Build, test and finalize a transformation.
Two-factor authentication (2FA): a 6-digit code from an authenticator app, asked for after your password. See Two-factor authentication.
Universe: the page that shows the DataMind product cards and the integration tiles. Open it from the sidebar. See Universe.
Upstream: in Lineage, the objects that the selected object reads from. The opposite is Downstream. See Explore and trace lineage.
View: a Warehouse object defined by a query. See Views.
View as: in DataPrism, runs every chart of a report under another viewer's row-level security, so that you can see what they see. See Row-level security and View as.
Visual: one chart, table, slicer, map or custom visual on a report page. See Visuals.
Warehouse: where Lake data is shaped into clean, modelled tables. See Warehouse.
Warehouse Builder: a Mind that builds Warehouse tables, views and transformations for you. See Warehouse Builder and Lake Builder.
Workflow: a named, scheduled group of loads for Lake, Warehouse or Writeback, with its own schedule, batch size and restart count. See Workflows.
Workspace: a container for decks, reports or models. A personal workspace is yours alone; a role workspace is shared by everyone who holds the role. See Workspaces, folders and your decks.
Writeback: sends data from the Lake or the Warehouse to external systems. See Writeback.
Writeback table: a table in a destination that receives rows from a Lake or Warehouse table, or from a SQL query. See Configure a Writeback table.
A Mind is the AI agent you chat with. An artifact is something a chat produced and saved, such as a query, a report or a slide deck. You find artifacts under Minds > Artifacts.
Deploy creates a Lake or Warehouse table physically, so that it can be loaded. A new Lake table needs a primary key, and it asks you to deploy it before you can open its settings, where Synchronize is. See Configure a table.
A sync type sets how a Lake table loads: for example, a full reload that truncates and inserts, an incremental upsert, a snapshot or a one-time load. You choose it on the table. See Choose a sync type.
In a Lake table's settings, no. Some labels there call the Lake table a staging table. In Observability > ClickHouse, Empty staging tables is about something else: the platform's working tables for loads and transformations, not your Lake tables. See ClickHouse storage and maintenance.
In Lineage, Upstream is what an object reads from, and Downstream is what reads from it. Use them to see what a change could affect. See Explore and trace lineage.
A console is one SQL tab in the Editor, with its own text, parameters and run log. Press Ctrl+Shift+Q to open a new one. Closed consoles come back through Find & reopen a closed console. See Consoles, saving and history.
RLS, or row-level security, limits which rows each user sees in a table. For Lake and Warehouse tables you set it in the RLS step of the table's popup. See Warehouse permissions and row-level security.
A workflow is the schedule and container for a set of loads. A job run is one execution of a workflow, and the runs table shows its status, start and end times and duration. See Job runs.
Universe is the page with the DataMind product cards and the integration tiles, such as the tiles for SQL clients, Excel and Power BI. Open it from the sidebar. See Universe.
The sidebar uses a capital B in WriteBack. The Dashboard cards and the rest of the product use Writeback. Both names mean the same area. See Writeback.