Bancheng / Setup time
If you're asking, you've probably already heard "first we'll schedule a discovery call, then we'll see about timing". This page explains what those weeks are spent on, which parts don't actually need you, and when we're slow too.
An enterprise AI chat project usually runs like this: discovery call, gathering content, loading it into the system, tuning the answers, internal testing, launch. Between each step someone waits for someone else's calendar, and those calendars aren't measured in days.
The slowest part isn't technical — it's gathering the content. Someone has to collect what's scattered across Word files, PDFs, an old website and your staff's heads into one document, and that someone is usually you. A consultant can ask the questions, but only you have the answers.
The knowledge base doesn't start blank. It grows from what you already have:
So the only thing left for you is putting up what needs saying. You'd have to do that anyway — it's your website.
There are 17 sites running on the platform right now. The fastest went from new account to live website in 56 minutes — not built by us, filled in by the business owner.
Since going live, in 89% of 655 real conversations nobody pressed the button to ask for a person. That number is customers voting with their behaviour, not us grading our own answers.
The fast side is easy to talk about; the slow side is the useful part. In these cases it takes longer, here too:
The free plan needs no credit card, and the assistant is ready to try as soon as you sign up. If you'd rather talk to someone first, that's fine too — we won't ask you to sign anything.