Tasks dashboard
New Tasks section in the platform portal. Shows AI jobs with filtering by customer, status, and date. The task detail panel includes a timeline of events with per-event cost and usage stats.
Changelog
Product updates, new features, and fixes. Updated at least twice a month.
New Tasks section in the platform portal. Shows AI jobs with filtering by customer, status, and date. The task detail panel includes a timeline of events with per-event cost and usage stats.
Two new profitability endpoints. An overview endpoint returns aggregated margin data across all customers. A time-series endpoint breaks down revenue, cost, and margin by period, event type, and category.
subscription_id is now included in credit grant API responses. The grants endpoint is also accessible directly from the platform portal.
Improvements across the Products section: product type labels, clearer price display, dimensional pricing ranges, and product version selection in customer subscription flows.
Webhook events can now be browsed in the platform portal. Filter by date range and select any event to open a detail flyout with the full payload and delivery status.
Customer rows are now clickable. External ID can be copied inline. Product version is shown when adding a customer.
The events list now includes a total costs column. Event rows are clickable. The date picker correctly highlights the current day.
Customer subscriptions now fall back to the account-level default success and failure redirect URLs when none are provided at subscription creation time.
Configure and manage webhook endpoint destinations from the Developers page in the platform portal.
Hybrid products now require at least one usage-based price. Attempting to save a hybrid product without a usage price returns a validation error.
MCP is now featured in the dashboard quick start guide with a dedicated setup banner.
Improved wallet adjustment form: options now show dates, validation runs on amount and reason changes, and asset display has been updated.
Two new API endpoints surface revenue and cost data for your customers. /customers/{id}/stats returns lifetime gross revenue, total costs, net revenue, and margin for an individual customer. /customers/stats returns the same breakdown across all customers, with a rolling 30-day summary by default (configurable up to 6 months).
Official Python SDK is now available. Documentation includes Python code examples alongside TypeScript across every API endpoint, with direct links to sample code on GitHub.
New /vendor/{id}/stats endpoint returns total costs attributed to a vendor, broken down by category. Filter by category, period_from, and period_to to scope the analysis. Useful for tracking spend per AI model or provider across your event stream.
Prices and entitlements on a product version can now be added or removed from the platform portal. Add fixed-price or usage-based prices, update entitlement limits and refresh strategy, and review a diff of all changes before saving.
The embedded customer billing portal has been refreshed with an updated layout and improved visual design.
The product page now shows the full version history. Switch between versions to inspect their configuration, mark a version as the default, or archive versions that are no longer in use.
Posting an event with an unrecognised vendor_external_id previously caused an API error. Unrecognised external vendor IDs are now auto-created so the event processes correctly — the vendor name can be updated afterwards. Supplying an unknown internal vendor_id still returns a validation error.
The events list now supports filtering by date range, event type, whether a fee was calculated, and event ID.
Official TypeScript SDK is now available. Documentation includes TypeScript code examples across every API endpoint, with direct links to sample code on GitHub.
The customer list in the platform portal now supports filtering by name, email, customer ID, external customer ID, and subscription status, making it easier to find customers in large accounts.
The active_subscription_count field was not being included in the customer API response. This is now fixed.
The Credyt MCP server now covers the full API surface. All documented API operations are available to MCP-connected agents, including endpoints that were previously missing from the tool list.
Issue one-off credit entitlements to any customer at the start of their subscription, useful for offering free credits on signup or trial extensions. Any products entitlements created without a refresh configuration will be one-off by default.
The redirect target after a customer's initial subscription payment is now configurable. When subscribing a customer, set redirect_to to portal for the Credyt portal or return_url for your own platform — useful when the post-payment flow lives in your product rather than the Credyt-hosted experience.
Published a set of AI skills that make integrating Credyt significantly faster. Four skills guide you from pricing strategy through to production integration. Install them via the npx skills CLI and they drop into any compatible agent setup.
The product catalog API now supports fetching the list of customers subscribed to a given product. Useful for building internal dashboards, running cohort-level operations, or syncing subscriber state to external systems.
Customers can now be queried by email address via the customer API, in addition to customer ID.
Credyt's API is now available as an MCP server. Connect it directly to any MCP-compatible AI agent or tool to manage your product catalog and customers in Credyt, and query billing activity without writing integration code.
Published the OpenAPI spec for the Credyt API and added support for webhook callbacks. Integrations can now subscribe to billing events and receive real-time notifications.
Added seat-based pricing as a billing model option. Teams and enterprise customers can now be charged per active user seat alongside or instead of usage-based charges.
Introduced outcome-based pricing, allowing charges to be tied to business results rather than raw usage. Useful for AI products where value is measured in completed tasks or conversions.
Free tier limits are now configurable per customer, cohort, or global default. Abuse prevention rules can be applied independently of paid tier thresholds.
Added support for B2B org hierarchy, allowing a single Credyt account to manage multiple sub-organisations. Usage and billing are tracked independently per tenant.
Issue one-off or recurring credit grants to customers, with configurable expiry and rollover rules. Grants are applied before paid balance and tracked separately in the audit trail.
Credit expiry behaviour is now fully configurable. Unused balance can expire on a fixed date or roll over indefinitely.
Resolved an edge case where duplicate usage events could be processed under sustained high throughput. Idempotency key checks now cover the full write path.
Added revenue vs cost correlation to the cost visibility dashboard. Each customer's billed revenue is shown alongside underlying model costs, making margin visible in real time.
Launched the embeddable customer billing portal. Drop a single component into your product to give customers live balance, usage history, and self-service top-up — no custom UI required.
Reduced live balance display latency from ~2s to under 200ms by moving balance reads to an in-memory store updated on every billing event.
Auto top-up is now available. Customers set a low-balance threshold and a top-up amount; their wallet recharges automatically without manual intervention.
Wallets can now hold multiple custom credit assets simultaneously. Each asset has independent balance tracking, pricing, and expiry rules.
Every usage event is now recorded with full token-level detail: model, prompt tokens, completion tokens, cost, and timestamp. Queryable per customer and per agent.
Native Stripe integration added. Connect a Stripe account to handle payment collection for wallet top-ups and subscription billing without building a custom payments flow.
Core billing engine shipped. Usage events are processed and charged in real time against customer wallet balances. A single POST to the usage endpoint is all that's needed to bill any AI action.