Pass an event
Your backend receives the incoming message and passes only the necessary context to the model.
INTEGRATION USE CASE
Give your AI a controlled communication channel
You choose the model and the assistant's operating rules. Senden API transfers messages, provides context through the API and accepts explicitly authorized reply commands.
View API →How to set up the workflow
Limit tools
Use OpenAPI, llms.txt and ai-context.json to describe commands. Keep the key in your backend, separate from prompts and the browser.
Review the reply
Check the result and business rules before sending. Provide for human handover and duplicate protection.
EXAMPLE WORKFLOW
Customer question → your backend → model and knowledge base → reply review → message in the original chat.
Integrate your chosen model into your own workflows without tying the transport to a single AI provider.
What to consider before launch
No built-in language model is provided. Account for prompt injection: an incoming message's content must not itself authorize sending, data access or arbitrary command execution.
Connect a test account, check the commands supported by the chosen messenger and run an agreed test in your project. Keep message IDs and use one idempotency key for each operation.
FROM IDEA TO INTEGRATION
Connect your first account
Create an account, verify your email and start by requesting the connection list.