Skip to content
Talk to our solutions team

Getting Started

The path from nothing to a running API. Each step is a command you can run.

If you just want to see how the model-driven part behaves without standing anything up, the Quickstart runs entirely locally and takes a few minutes.

kis.ai runs on your infrastructure, so you need a licence before you can install it. Talk to the solutions team to get one, then follow Install to set up KVM and the CLI.

Everything below goes through kis, installed via KVM — see Install for the full flow. Confirm it is present:

Terminal window
kis version

kis --help lists what you actually have; the command list is assembled from the bundles installed on your machine, and commands you are entitled to but have not installed are listed separately with the bundle that provides them.

Terminal window
kis login

This obtains a token for your tenant. Most commands need it: the platform is multi-tenant and every request carries tenant identity.

A product is the unit you build and deploy — its blocks, entities, configuration and code.

Terminal window
kis info products

Meta holds what a product is made of; Config holds its settings and service discovery. Those two come up before anything else is meaningful.

This is where the model-driven part starts paying off. Declare an entity in YAML and the Data API serves REST and GraphQL over it — no endpoint code to write.

Terminal window
kis spec validate # check YAML against the schema
kis spec lint # style and consistency

Install the IDE language server first if you have not. It validates the same schemas as you type, which is a far shorter feedback loop than running spec validate after the fact.

Configure blocks in YAML rather than writing services:

NeedBlock
Login, roles, tokensIAM
Entities, REST, GraphQLData API
Secrets and certificatesVault
Model access with budgetsAI Gateway
Search and RAGAI Knowledge
Multi-step workAI Flow
Background jobsJobs
Email, SMS, pushNotifications

Start with the smallest set that makes your first screen work. Blocks compose, so adding one later is configuration rather than migration.

Terminal window
kis test -t tests/

Test definitions are YAML that lives alongside your code. See Testing for the definition format, and Code Quality to scan the source before you ship it.

Terminal window
kis info services # what is deployed, and its state

Supervisor keeps declared services in their desired state on a node, Observability collects the telemetry, and Automate runs the pipeline that gets you there.