The questions we actually get.

You do. The bot runs on your keys and in your accounts: access to the model, the CRM, and the messaging channels is set up under your company. We get access to configure it, and you can revoke it at any time without asking us.

This is the first question people ask on calls, and rightly so. If a contractor keeps the keys to themselves, you're not paying for the agent — you're paying for the right to use it for as long as they let you.

Everything stays with you: access, conversations, settings, and the accumulated knowledge base. We don't move the bot onto our own infrastructure or tie it to ourselves. Anyone who understands your CRM will be able to configure it further.

In your CRM, together with the deal card. There's no separate place where we store your conversations and you never look.

The cost consists of two parts: a one-time setup fee and a monthly fee for operation. The monthly fee depends on how much knowledge the agent holds: if the materials fit into the prompt, it's cheaper; if full search across catalogues and price lists is needed, it's more expensive.

We give the exact figure after the first call, once it's clear how much material and how many channels are involved.

It depends on the volume of knowledge, the number of communication channels, and how much the agent does on its own. Answering common questions costs one thing; calculating prices by size and maintaining a CRM record cost another.

Besides the setup, there's a cost for the model itself: it's charged by conversation volume. For a business handling dozens of enquiries a day, this usually works out noticeably cheaper than an hour of a manager's time spent handling them by hand.

Quote an exact price without an on-site assessment, promise deadlines, offer discounts, or make decisions on your behalf. Boundaries are set at the start, and this is a separate conversation: every business has its own list of things only a human should say.

It answers based on your materials, and passes on anything not covered there to a human. In the first weeks after launch, conversations are reviewed in full — this is how spots where it misunderstood a question are found and closed.

Yes, and that's usually the main reason it's deployed. The agent finds out what the client needs and tags the enquiry with a category — from "call this one today" to "not our client." The manager stops spending time sorting incoming leads and focuses on those ready to buy.

No. It takes off the manager's plate the things done by hand and without enjoyment: the first reply, clarifying questions, sorting through incoming messages. Conversations that need a human stay with a human.

WhatsApp, Telegram, the chat on your site. An enquiry from any channel lands in the CRM as one record, not three.

WhatsApp has a 24-hour window: if the client hasn't replied, you can't message them again without a separate template. Telegram has no such limit, so long nurturing sequences are usually built there.

No. The agent is built into the one you already have. If there's no CRM at all, that's a separate piece of work, and we'll say so directly rather than bolt the agent onto nothing.

Whatever a new manager would learn from: catalogues, price lists, answers to common questions, work rules. If it's scattered across ten files — send it as is, bringing it together is our job.

Send them as they are. The agent works more accurately with a spreadsheet than with a scanned catalogue, so we'll convert the materials into that format ourselves.

The first working version takes days, not months. It answers questions and qualifies the enquiry. What comes next is added based on what's visible from real conversations — and that's the right order: launch first, add complexity later.