Deploying: Embedded Chat or Exported Code

Hosted workflow or your own runtime

A published workflow can be served through OpenAI's hosted runtime and embedded in your product with ChatKit, a customisable chat interface that your backend connects to the workflow (your server creates sessions so API keys never reach the browser). Alternatively, export the workflow as Agents SDK code and run it in your own service, where you control tools, data and deployment. Either way, authenticate users in your backend, pass trusted identifiers as inputs, and set limits on usage per user. Product details such as node names and menus change; check the current OpenAI documentation.

Deployment options compared

Choose by control and effort.

                    embedded chat (hosted workflow)      exported SDK code
runs on             OpenAI platform                      your servers
UI                  embeddable chat component            your own UI
custom tools        via MCP / hosted tools               any Python/TypeScript code
changes             publish new workflow version         code review + deploy
best for            fast launch of a chat experience     deep integration, strict control

Keep keys server-side

Never put API keys in browser code; issue short-lived session tokens from your backend.

त्वरित जाँच: Why create chat sessions from your backend?

• So API keys stay on the server and users are authenticated by your app
• Browsers cannot display chat
• It makes the model smarter
• It removes the need for versions