Loading
Because they fail in the two available ways. Scripted trees cannot answer anything real, so visitors type "agent" immediately; general models answer everything confidently, including the things they have invented.
The Pulse chatbot answers visitor questions from the sources you trained it on, citing the source for each answer.
It answers only from your trained sources and links the one it used. Confidence is not enough — an answer it cannot cite is not sent, and the conversation goes to a person instead.
A visitor asks whether their contract covers weekend support. There is no source for that customer's terms, so the bot says so and offers a person rather than describing a typical policy.
It will never commit to a refund, a discount, a date or an exception, and it never claims to be human. Both are rules rather than settings.
Your trained sources, and — for a signed-in customer — their own record. Never another customer's anything.
No, and that is deliberate. No sources means no answers.
Yes, from the same sources, in the visitor's language.
Yes, with the sources used, reviewable at any time.
Fourteen days, every module, no card. Or half an hour with someone who will run it on your own records and tell you where it does not help.