Put a cited chat on your website: visitors get answers from your price list, product sheets and FAQ — around the clock, with a link to the source, and an honest «I don't know» when the answer isn't there.
The examples on this page are illustrations — the answers come from your own documents, not from ours.
Most questions a website gets are already answered somewhere on it. Opening hours, delivery terms, what a service includes, which product fits which need. The visitor just cannot find it — so they write an email, call, or leave.
A generic AI chatbot does not solve that. It answers from the internet, sounds confident, and now and then invents a price or a policy you never had. For a business, a wrong answer with your logo on it is worse than no answer.
You add your own material to a brain — the price list, product sheets, terms, the FAQ, the pages already on your site. Trail reads it and builds a knowledge base from it.
Then you place the chat on your website with one line of code, the <trail-chat> element. Visitors ask in their own words and get an answer built from your material, with a link to the source so they can read the details themselves.
When the answer is not in your material, the chat says so — and can point the visitor to a person instead. That is a deliberate choice: a chat that admits a gap keeps your customers' trust; one that guesses spends it.
Every question is also a signal. You can see what visitors ask and where the brain had no answer, so the next document you add closes a real gap instead of a guessed one.
in the inbox and on the phone.
The first conversation starts where the website left off.
— from your own terms, not from the internet.
Change the price list in the brain, and the chat follows.
Every answer shows which documents it is built on.
New knowledge waits in a queue until a person approves it.
When two sources say different things, someone at your company decides.
Trail is hosted in Stockholm, Sweden, and answers are written by an EU language model.
Most businesses start with the material they already have: the FAQ, the price list and the terms. Send us those, and we will show you your own website answering from them.
Or see it first — the demo asks a brain the question «What did visitors ask the chat on our website this week, and which questions could it not answer?» and shows the answer with its sources.