Core concepts

Klarety has a small set of building blocks. Each one below links to the section that covers it in depth.

How the pieces fitCopy Anchor Link

A workspace holds everything. Inside it, members work on projects. Each project has a Studio where code is written, with AI alongside. Data lives in Nexus, your database, as tables grouped into schemas, and as files in object storage. Connectors bring data in from other tools. Agents do recurring work on that data. Deploys put a project online. All usage is paid for with credits.

WorkspaceCopy Anchor Link

A workspace is your team's home in Klarety, with its own projects, data, members and plan. Nothing is shared with your other workspaces. You can belong to several and create new ones from Settings. Its name, logo and Workspace ID are under SettingsWorkspace.

See Workspace settings.

Members and rolesCopy Anchor Link

Members are the people in a workspace. Each has one role:

• Owner and Admin: run the workspace. Only they change the plan, manage API keys and take agents live.
• Member: works in the projects and data they have been given.

Owners and admins always reach all data. For everyone else, Data access decides what each person can see and change, from knowing a table exists up to changing its data or granting that access to others.

See Members and Data access.

ProjectCopy Anchor Link

A project is one piece of work: an app, a set of analyses, a data pipeline. It holds your code (optionally linked to a GitHub repository), secrets, settings, agents and deploys. Project people: the owner can do everything, an Admin can open the Studio, change settings and manage people, and Use can open the Studio.

See Create a project and Project settings.

StudioCopy Anchor Link

A Studio is a project's cloud workspace: code, a terminal, a live preview and AI that can change the project. You start, open and stop it from the project. Several people can work in one Studio at once, each with their own AI session.

A Studio runs on a standard shared machine by default. If that machine is taken back, a new one starts on its own and no work is lost. Premium machines are never interrupted.

See The Studio and the AI harness.

Nexus and the Nexus engineCopy Anchor Link

Nexus is your database, where tables live and are queried. In the app it is called Nexus Lakehouse. Tables are stored in the open Apache Iceberg format, so your data is never locked into a format only Klarety can read.

The Nexus engine runs your queries from the SQL editor, agents and code. Developers can also open a Spark session on it (Spark SQL and Spark Connect).

See Data and Spark.

Tables and schemasCopy Anchor Link

A table is rows and columns, like a spreadsheet that can grow to any size. Tables keep their history, so you can see what changed and go back. Checks set aside rows that do not fit, and column descriptions help people and AI understand the data.

A schema is a group of tables, such as sales or marketing. Access can be given to a whole schema or to one table.

See Tables and Undo and restore.

Object storageCopy Anchor Link

Object storage holds files of any kind, in folders: PDFs, images, recordings, spreadsheets. Every version is kept. A spreadsheet can become a table with one action, and a folder can import new files automatically. Deleted files go to the trash first.

See Object storage.

Knowledge graphCopy Anchor Link

The knowledge graph shows what your data is about, such as customers, products or places, and how they relate across tables. You can browse its types and records and find groups.

See Knowledge graph.

ConnectorsCopy Anchor Link

Connectors bring data in from tools such as GitHub, Google Drive, PostgreSQL or Amazon S3. GitHub connects directly; for the others, your Studio's AI sets up the connection. Connector credentials are encrypted.

See Connectors.

Agents and runsCopy Anchor Link

An agent is a helper that works on its own. You describe the task in plain words and the AI builds it as a draft. It starts on a schedule, when data arrives, when a webhook calls, or when you press Run, and asks before changing anything important.

Each time an agent works is a run, showing every step, what it produced and what it cost. A draft does nothing until a workspace admin takes it live.

See Agents and Runs.

DeploysCopy Anchor Link

A deploy puts a project's app online at a free address, and you can add your own domain. You choose how it runs (sleeps when idle, or always on), and you can roll back to any earlier deploy or take the app offline.

See Deploy an app and Domains.

CreditsCopy Anchor Link

Credits pay for Studios, live agents, deployed apps and domains. One credit equals one US dollar. Your plan includes monthly credits, and you can buy more. Balance, spend and limits are in SettingsUsage & limits.

See Credits and Billing.

Next stepsCopy Anchor Link

• Quickstart: put these pieces to use in a few minutes.
• Security and trust: how your data, code and keys are protected.
• Your first project: a full walkthrough of a project and its Studio.