Build on the platform every customer runs through
REST APIs, real-time events, workflows, installable apps, chat SDKs and AI tooling for Kustomer. Detailed enough for engineers, structured enough for the models working beside them.
$
From zero to a live integration
Three steps cover most integrations: authenticate, call the API, and listen for changes. Each links to the full guide.
Create an API key
In Settings → Security → API Keys, pick the narrowest roles your integration needs.
export KUSTOMER_API_KEY=eyJhbGciOi…Read the guide
Make a request
Call your org-scoped host with a Bearer token. Responses are JSON:API-style envelopes.
curl https://acme.api.kustomerapp.com/v1/users/current \ -H "Authorization: Bearer $KUSTOMER_API_KEY"Read the guide
React to events
Subscribe an outbound webhook to conversation and customer changes, and verify each delivery.
{ "events": ["kustomer.conversation.update"], "url": "https://example.com/hooks" }Read the guide
Everything you can build on Kustomer
Concepts first, then recipes, then the reference. Every page links down to the exact endpoints it uses.
Docs your agents can read
Every page, endpoint and schema is available as structured text, so coding assistants answer from the source instead of guessing.
- Markdown for every pageAppend .md to any URL, or use Copy page. Clean Markdown, no navigation chrome.
- llms.txt and llms-full.txtA curated index and the full corpus, following the llms.txt convention.
- Kustomer MCP serverConnect Claude, ChatGPT or Cursor to your Kustomer org with OAuth sign-in. Each person sees only what they can in Kustomer.
# Create customer> POST /v1/customers · Core resources API## Required rolesAny one of: `org.permission.customer.create`## Request body- `name` (string, nullable)- `emails` (array<email>)- `externalId` (string, nullable)
Kustomer chat, native everywhere
One conversation model across web and mobile, with identity verification, push notifications, knowledge base and AI assistants built in.
Ship your first integration today
- Every operation, with request and response schemas
- Samples in cURL, Node.js, Python and Go
- A live console on every endpoint
- Markdown twin of every page