Getting started

Core concepts

Projects, datasets, ontology objects, links, actions and apps.

Projects and resources

Work in Converge lives in projects. Datasets, agents, functions and other resources belong to a project, and access is granted on the project or the resource. A new resource, such as an agent you create, starts private in your personal project until you share it.

Resources carry markings. A marking restricts who can see a resource's contents. See Markings and classification.

Datasets

A dataset holds tabular data or files. Datasets have branches, and each change is a transaction (append, update, delete or snapshot) that is opened, written and then committed or aborted. The schema is versioned per branch, so you can see how a dataset changed over time.

Datasets are filled by syncs from external sources, by file uploads, and by pipelines.

Ontology objects

An object type describes a kind of real-world thing: its properties, its primary key, and the dataset that backs it. Each row becomes an object that applications and agents can search, filter and aggregate.

Links

A link type connects two object types, for example an Order to its Customer. Links have a cardinality (one-to-one, one-to-many, many-to-one or many-to-many) and can be followed in both directions.

Actions

An action type is the governed way to change objects. It declares parameters, the edits it makes (create, modify, delete, link), who may run it, and optional side effects such as notifications or webhooks. Some actions require approval before they take effect.

Functions

A function is a small read-only computation over objects, published so that applications, actions and AI logic can call it. Functions run as the caller, so they see only what the caller may see.

Apps

Apps are the interfaces people use: built-in applications (Objects, Map, Object Analytics, Approvals) and custom apps built on the ontology. They all read and write through the same APIs.