An assistant that answers your customers from your documents, not from what a model believes it knows about your trade.
A language model on its own knows neither your prices, nor your lead times, nor your terms. Shipped as is, it answers anyway: that is the main flaw of these tools, and the one that costs the most in lost customers.
So we connect the assistant to your own content and hold it there. When the answer is not in it, it says it does not know and offers to put you in touch. That sentence is worth more than an invented answer.
Most of the messages a small business receives are the same five questions. Opening hours, lead times, coverage area, rough price, how to start. Answering them around the clock frees time without removing the human contact, which stays for what matters.
The assistant can also prepare the work: qualify a request, summarise it and file it, so it reaches you already sorted.
Conversations are processed by a model provider, and that is a choice made with open eyes. We tell you which one, where its servers are, what it retains and for how long, before anything is signed.
By default we pick providers that process in Europe and do not reuse your exchanges to train their models. If your business cannot send certain data to a third party, we say so beforehand, not afterwards.
