Developers
Building on Kinetic
Kinetic is where you build applications and workflows over the ERP, HRIS and ticketing systems you already run. Forms, workflows and integrations live in a space you change through the console, the Core REST API or an AI client over MCP. Every definition exports as JSON or XML you can keep in Git.
Core REST API GET /app/api/v1/kapps lists every app in your space. Documented in the API Reference for version 6.1. API quick start
Architecture
Three layers on top of the systems you already run
People and agents work through the Experience, Workflow and Integration layers, which read from and write to your systems of record. Nothing is migrated into Kinetic unless you choose to store it there.
Where the experience layer lives
- Portals
- Front-end bundles you write (usually React), host where you choose, and connect to Kinetic over REST.
- Forms
- Defined and stored in Core. Conditional logic and custom JavaScript run in the browser; bridges supply live data.
Where the workflow layer lives
- Task (workflow engine)
- Runs workflow trees and reusable routines when a form is submitted or an event fires.
- Handlers
- Ruby functions that workflows call for custom logic and system calls.
Where the integration layer lives
- Integrator
- Connections and Operations for REST and SQL systems, with credentials held apart from users.
- Kinetic Agent
- Installed inside your network. Runs bridge adapters so forms and portals can reach systems behind a firewall.
Deployment options and extension points are on the technical architecture page.
Ways to build on Kinetic
All three are available. A typical project uses an AI client for the first draft, the REST API for traffic from other applications, and integrations to reach the systems of record.
-
Build with AI through MCP
AvailableConnect Claude Code, Cursor or another MCP client to a development space. The AI builds forms, workflows and integrations as the user it signs in as, and a person reviews the export before it is promoted.
How the MCP server works -
Call the Core REST API
AvailableRead and write spaces, Kapps, forms and submissions over HTTPS. A submission made through the API starts the same workflow as one made by a person.
Authentication and quick start -
Extend the integrations
AvailableReuse published handlers, bridge adapters and Connections. If a system has an OpenAPI spec, a Connection can call it; for anything else, write a handler in Ruby or an adapter in Java.
How integrations are built
Working example
Submit a form and read the result over REST
List the apps in a space, submit a form, read the submission back. Submitting starts the workflow attached to the form.
- Base URL. Every Core call goes to
https://<your-space-url>/app/api/v1, scoped to one space. - Authentication. HTTP Basic with a space user, as the API Reference specifies.
- Permissions. Responses only include what that user's security policies allow.
Endpoint details: Kapp Search (opens in a new tab), Submission Create (opens in a new tab) and Submission Retrieve (opens in a new tab).
# 1. List the Kapps (apps) in your space.curl -u "$KINETIC_USER:$KINETIC_PASSWORD" \ "https://<your-space-url>/app/api/v1/kapps" # 2. Submit the "new-hire" form in the "services" Kapp.# completed=true submits every page; workflows on the form start.curl -u "$KINETIC_USER:$KINETIC_PASSWORD" \ -X POST "https://<your-space-url>/app/api/v1/kapps/services/forms/new-hire/submissions?completed=true" \ -H "Content-Type: application/json" \ -d '{"values": {"Employee Name": "Jordan Lee", "Start Date": "2026-11-02"}}' # 3. Read the submission back, with its values and details.curl -u "$KINETIC_USER:$KINETIC_PASSWORD" \ "https://<your-space-url>/app/api/v1/submissions/<submission-id>?include=values,details" Tested on version: pending verification
The docs are the source of truth
Procedures and API definitions live at docs.kineticdata.com. The current version is 6.1; 6.0 and 5.2.0 remain for existing installs.
- Platform Guide (opens in a new tab)
Procedures, from your first form to a customer-managed install.
- API Reference (opens in a new tab)
The Core REST API as an OpenAPI specification.
- Releases (opens in a new tab)
Release notes for 6.1, 6.0 and 5.0.
- Integrations Catalog (opens in a new tab)
Every published integration artifact, by system.
- llms.txt (opens in a new tab)
A plain-text index for AI agents. Every docs page also has a markdown version.
Where to go next
-
Reference builds
One reference build per solution area, published once each has been run against a real tenant.
Open -
Agent-native work
The thesis for agents and people sharing one process, and how to join early access. In development.
Open -
Technical architecture
What each service stores and runs, and where you can deploy it.
Open
Questions developers ask
Answered by Kinetic Data. Last updated .
-
Is this low-code lock-in, or can I keep configuration in version control and ship through my own pipeline?
Kapps, forms, workflows and WebAPIs export as JSON or XML, from the Space Console or the API, so they can live in Git and be imported by your pipeline. Custom code stays in your repositories: portals are front-end bundles, handlers are Ruby and bridge adapters are Java.
-
What exactly does the MCP server expose, and what can AI not do?
The Core and Integrator APIs, as tools such as forms, workflows and submissions. AI mostly builds with it, and can also create records and start workflows. It cannot exceed the signed-in user's security policies or change how a workflow runs once started.
-
How are AI-generated changes reviewed before they run?
Like any configuration change. Build in a development space, export and diff the definitions in version control, then import them into production. Versioned workflows let in-progress runs finish on the version they started with.
-
Which authentication methods and deployment models are supported?
SAML or OIDC single sign-on (Entra ID and Okta included), LDAP, CAC/PIV and internal accounts; the Core REST API takes HTTP Basic. Kinetic runs in Kinetic Data's cloud or customer-managed on Kubernetes, including air-gapped.
-
How do I call Kinetic from my own agents, and what is coming for agent-native work?
Through the MCP server or the Core REST API: create a submission to start work, read it back for status. Assigning workflow steps to agents the way they are assigned to people is in development.
-
Where do I run it, and how does it behave in an air-gapped environment?
In Kinetic Data's cloud, or on Kubernetes in your own cloud or data center. Air-gapped installs pull images from a private registry and are upgraded offline.
Build one real process with our engineers
Bring a process that crosses two or more systems. In a free two-hour session we map it to forms, workflows and integrations in a Kinetic space and show you the API calls and configuration behind it.