AI enquiry triage
AI triage for incoming enquiries
Enquiries arrive by e-mail, through website forms, in chat and as voice messages, in several languages and in no particular order. The model extracts what matters, fills in the client record, drafts a reply and creates a task with a deadline. Reading it, correcting it and sending it stays with your team.
At a glance
- Who it is for
- companies with a steady flow of enquiries across several channels and languages
- What is included
- message processing, CRM record, draft reply, task with a deadline, processing log
- Price
- from €1,500; language model usage is billed separately, according to actual consumption
- Timeline
- usually 2–4 weeks
- Control
- the AI prepares drafts and fills in fields; a person decides and sends
- Languages
- the model understands Slovak, Russian, Ukrainian and English, and replies in the client's language
How it works, step by step
- A message arrives by e-mail, from a website form, from chat or as a voice message. Voice messages are transcribed into text first.
- The model identifies the language of the message and reads its content.
- It extracts the essentials: name, contact details, field, request, budget and urgency. If something is not in the message, the field stays empty and flagged — nothing is guessed.
- It fills in the client record in your CRM and assigns it to the right business line and source.
- It drafts a reply in the client's language and in your tone of voice, following the rules we agree with you.
- It creates a task for a named colleague with a deadline, and for urgent messages adds an alert in the team chat.
- Your colleague reads the draft, edits it and sends it. Only then does the client receive a reply.
What the person does and what the AI does
- AI: transcribing voice messages, detecting the language, extracting data, filling in fields, drafting the reply, proposing a deadline
- Person: checking the data, editing the text, prices and commitments, the decision and the sending
- Where a mistake is expensive — prices, deadlines, legal or medical content — we add a mandatory review before anything is sent
- The rules are written together with you: what may be filled in automatically, what may only be suggested, and what must never be sent
- When the model is unsure, it leaves the field empty and flags the enquiry for manual handling — empty is better than invented
Data and the processing log
The model receives only what the task requires: the text of the message and the fields it has to fill in — not your whole client database. Which data is processed, where it goes and on what terms is set out in the contract before work begins, and the system runs on hosting inside the European Union. Every run leaves a log: what the model received, what it returned, who approved it and what was finally sent. That makes it possible to go back, see why a reply looked the way it did, and adjust the rules.
About the demo on our home page
The demo on our home page is a scripted walk-through with a fictional client, and it is labelled as such. It is not live processing and it is not a real enquiry. Its only purpose is to show the order of the steps: message → record → draft reply → task. The live model runs on the client's own project, on their data and under their rules. Before launch we test it on your own past messages, and you review the results before anything touches day-to-day work.
What you get and what it costs
Implementing this single scenario starts at €1,500 and usually takes 2–4 weeks. The first consultation lasts 30 minutes and is free of charge; after it you receive a fixed quotation on a single page. Language model usage is billed separately, according to actual consumption — it is a line of its own, shown apart from the cost of our work. There is no monthly fee per user, and the code and the data are yours. If you have no CRM, the scenario can be connected to e-mail, Telegram or the system you already use; a bespoke CRM is a separate service, from €3,000.
Questions and answers
What does it cost?
Implementing the scenario starts at €1,500 and usually takes 2–4 weeks. On top of that you pay for language model usage according to actual consumption; it is metered separately and shown as its own line. The first consultation lasts 30 minutes and is free, after which you receive a fixed quotation on a single page.
Does the AI send replies on its own?
No. The AI prepares the draft and fills in the fields; a person decides and sends. Where a mistake would be expensive, we add a mandatory review before sending.
What if the model misunderstands something or invents it?
The rule is simple: what is not in the message is not filled in. The field stays empty and the enquiry is flagged for manual handling. Every run is logged — what the model received, what it returned and who approved it — so a mistake can be traced and the rule corrected.
Is the demo on your site a real client?
No. It is a scripted walk-through with a fictional client, and it says so. It shows the order of the steps, not live processing. The real model runs on the client's project, on their data and under their rules.
Do we need a CRM for this?
No. The scenario can be connected to e-mail, Telegram or a system you already use. It is more useful alongside a CRM, because the result immediately becomes a client record and a task with a deadline. We build bespoke CRMs separately, from €3,000.
Other services
Updated: 2026-09-19
Tell us what you want to fix or build
The free discovery call takes 30 minutes. We reply within one working day and send a fixed quote on a single page.