Kaiten

Introduction

What's Coming

What Kaiten is building next — the commercial layer on top of Licenses first, then the rest of the roadmap — and the direction it serves.

Licenses, entitlements, feature flags and deployments are the model in production today. What comes next extends that model: a price attaches to a License, a voucher acts on a grant or a flag, an offline agent serves the same entitlements.

Roadmap, not reference. Nothing on this page is in the product yet. A capability is available once it appears in the API reference and its row in the editions matrix no longer carries Coming soon. The public roadmap tracks each item and takes votes.

In progress

The commercial layer on top of Licenses:

  • Native billing — Kaiten rates each Instance against the prices of its License; the billing engine you already use invoices and collects.
  • External billing engine adapter — a provider-agnostic adapter that maps your catalog onto that billing engine: one product per License, one price per License price.
  • Add-ons — stackable extensions that raise or override a License's grants, without forking the License.
  • Vouchers — redeemable codes that act on a price, a grant or a flag: a discount on one usage price, a temporary quota increase, a feature unlocked without a deploy.
  • AI credits — models and conversion rules on top of the existing NUMBER_AI_CREDIT entitlement: per License, you set how many credits a request on each model costs. This is the core of the AI Control Plane.
  • Threshold behaviors — a grant enforces a hard cap or a bounded allowance today; a grace period and a notify-only mode join them, set per License with overrides per Instance.

For developers and agents:

  • MCP server and Skills — every capability ships with a Skill, so an agent can create Licenses, define entitlements, configure flags and provision Instances.
  • React components — a pricing table, a customer portal and a paywall, wired to the API.
  • Your own identifiers everywhere — Customers carry your identifiers today; Instances, entitlements and deployment zones get them too, so a CRM deal or a billing subscription maps onto Kaiten without a reconciliation script.

Planned

  • Kaiten Edge — an agent inside air-gapped or self-hosted environments that serves flags and entitlements with no connection to Kaiten; usage returns as signed bundles that Kaiten verifies.
  • Feature entity and lifecycle — each product capability becomes an object that moves from planned through beta to general availability and deprecation, linked to the flags that roll it out and the entitlements that sell it. Its history outlives the flags.
  • Feature tracking — send your product events as they are; filters match them to features and turn them into usage, with no separate metering call.
  • Cohorts and configuration — group Instances by hand, by rule or both, and give them typed settings that cascade from a global default to the cohort to the Instance.
  • Roles and ownership — built-in roles made of the scopes you already know, and an owner on each Customer and Instance, so the whole revenue team can work in the console.
  • Code generation — point the CLI at your catalog and get typed Licenses and entitlements in Go, TypeScript or Python.
  • Composable pricing — several prices on one License: a base fee plus one usage price per metered entitlement.
  • Per-user usage — attach your own opaque user identifier to usage and events, and see consumption per end user inside each Instance.
  • HubSpot connector — the CRM sync the Attio connector provides, for HubSpot.
  • Activity alerting — rules on real usage, such as an Instance gone quiet, a sharp drop, or a quota close to its limit month after month, that reach your CRM as churn and upsell signals.
  • Marketplace deployment — define Licenses and entitlements once; Kaiten prepares the AWS Marketplace listing and provisions the Customer, License and Instance on each purchase.
  • Embedded AI assistance — threshold suggestions from observed usage, alerts on consumption anomalies, help with pricing.

Further out

  • System map — what runs where and for whom, drawn from the components, releases, deployment zones and Instances Kaiten already holds, with what a change would affect.
  • Public changelog — release notes for each Customer, from the releases deployed on their Instances.
  • Workflow engine — Kaiten's events turned into durable automations, with native actions on Licenses, entitlements and Instances.
  • Customer intelligence — a health view of each Customer, built on usage, flag activity and License history.
  • Data exporter and SQL explorer — the audit trail and usage exported to your data warehouse, and queried from the console.
  • Skills registry — Skills contributed by the community, by framework and use case.

The direction

Kaiten's premise is that the commercial contract is the source of truth of the infrastructure: one model for what a Customer bought, what they can access and what runs for them, read the same way by your services, your teams and your agents. Each item above adds to that model.

Kaiten connects to the tools that already do their job — billing engines, CRMs, deployment runners, LLM gateways, data warehouses — and leaves that job to them.

Where to go next

What runs today: Monetization and the editions matrix. The terms that arrive with billing: Vocabulary.

On this page