Operational apps
Operational apps: overview
What an operational app is on Quantum, how it runs, and what you get without building it: history, permissions, files, executions, and analytics.
Updated: 2026-11-17
In this guide
What an operational app is
An operational app is a day-to-day work application (returns, cash closures, reconciliations, cases) that you do not code from scratch. You describe it as data and Quantum runs it.
You bring two things:
- A definition: a JSON document that declares the objects, actions, permissions, files, and screens.
- Flows: the code with your business logic. They run outside the definition, when an action or a file triggers them.
Quantum brings the rest: storage, validation, access control, history, file uploads, flow dispatch, and the interface in the dashboard.
The pieces
- Record types: the objects of your app. Each type has a field schema and, optionally, a lifecycle with states and allowed transitions.
- Records: the instances. Each record keeps its data, state, version, and history.
- Actions: what a person can do. An action can ask for input, write fields, move the record to another state, and start a flow.
- Flows: your logic. They receive a standard envelope with the record, the action input, and the app settings, and return results by writing records or dataset rows.
- Permissions: roles with capabilities, assigned to users or to organization roles, with optional scope (for example, by store).
- Files: slots where files are attached to a record or to the app, with type, maximum size, and integrity verification.
- Datasets: tables of high-volume rows that flows write and screens read.
- Views: the screens, composed from ready-made components (tables, metrics, detail, actions, files, charts) and, when needed, your own components.
How it runs
- You store the record types and the app definition in your organization.
- Quantum validates the whole definition. If anything references a piece that does not exist, it is rejected and every error is listed.
- A user opens the app in the dashboard. They only see what their permissions allow.
- The user runs an action. Quantum checks permission, scope, and state, validates the input, writes the record, stores the history, and queues the flow, all in one operation.
- Your flow starts with the standard envelope, does its work, and registers the result.
- The screen refreshes on its own with the new state.
Your definition is data. Your logic is flows. Quantum executes the contract and does not need to know your use case.
What you get without building it
- History: every change stores who, when, which fields changed, and a copy of the record. Changes made by an action also store the action and the input sent.
- Permissions: capabilities by role and scope by dimension, enforced on the server on every read and every action.
- Safe concurrency: every record has a version. If two people act on the same record, the second one gets a conflict instead of overwriting.
- Verified files: uploads go straight to storage with a SHA-256 checksum. A file only counts once Quantum confirms the content matches.
- Executions: a flow bound to an automation shows up in Executions like any other run, with the user who triggered it.
- Exactly-once start: the flow start is stored together with the record change. Retrying the same request does not start the flow twice.
- Analytics projection: every version of every record is copied to Quantum's analytical store for reporting.
- Environments: the same definition is installed separately in QA and in production.
Records or datasets
Use records when the object:
- Has a lifecycle (requested, approved, closed).
- Receives actions from people.
- Needs history, attached files, or scoped permissions.
Use datasets when the data:
- Is many rows produced by a flow (statement lines, matching results, sales detail).
- Is only read, filtered, and charted.
- Can be rewritten in full on every run.
A practical rule: if someone will make a decision about the row, it is a record. If the row only explains or supports a decision, it is a dataset.
Keep in mind that datasets have no per-row scope: anyone who can open the app can read every row of its datasets.
Next steps
- Quickstart: build a returns app end to end.
- Definition reference: every field of the document.
- Flows: how to write the logic.
- Views: how to compose screens.
- Permissions and environments: roles, scope, QA, and production.