EnterChat — A WhatsApp assistant your staff can teach

Documentation

How EnterChat works, what it measures, and how to drive it from your own systems.

Integration guide

Connect a WhatsApp number, teach it from files you already have, and release it one intent at a time.

Every endpoint with example requests and responses. Runnable against this deployment.

Did anyone reply?

Every message answered in seconds, in the customer's language, from your price list and your hours — never invented.

Is it getting better?

Drop in last year's chats, correct a wrong answer in one click, or rehearse with it yourself. Every correction is a visible, reversible lesson.

Did it turn into work?

A question becomes an appointment in the same thread, and everything it could not close is logged as a lead with the conversation attached.

Are you still in charge?

Each kind of question earns its own autonomy. Prices answer themselves; complaints never leave the building without you.

Service window: The 24 hours after a customer's last message, during which replying is free. Once it closes, reopening the conversation costs a paid template — which is why the inbox counts down to it.
Deflection: The share of conversations closed without a person. Not the same as being right — it is the share nobody had to touch, which is why accuracy is tracked separately per intent.
Autonomy state: How much rope one kind of question has earned: draft only, suggest, or auto-send. It is per intent, not per assistant, and it moves in both directions.
Lesson: One correction, stored as an object you can read and undo — what it used to answer, what it answers now, and which conversation taught it.
Knowledge gap: A question customers keep asking that nothing in the knowledge base answers. Mined from escalations and repeat contacts, ranked by how often it comes up.
Topic map: What an import produced: the topics found in your old conversations, how customers phrase each one, and a proposed answer waiting for you to approve, edit or throw away.