To qualify a prospect, ask them what you need to make a fair proposal, not a generic checklist. When you create a Muzau agent, you choose the list of details to collect, ten at most, suggested from your line of work. The agent then asks each prospect who writes in, files their answers in their own words, column by column, and tells you when everything is in. You read the next step before it goes out.
A prospect writes three lines: they are looking for something, and they would like a price. To answer them properly, you almost always lack one detail: the quantity, the deadline, who decides, what they use today. You ask, they answer half of it, and the meeting gets booked without knowing what they are really after.
Qualifying is not giving an exam. It is gathering, before you commit time, what lets you make the right proposal, or say no early and politely. Your team keeps the judgment: who to meet, what to offer. Asking the questions, following up and filing the answers take time, and that is the part the agent takes on.
When you create an agent that qualifies, Muzau suggests the list of details to collect based on your line of work. You remove some, add some, ten at most: they become the columns of its tracking file, and the questions it asks each person who writes in. A detail marked "as applicable" is noted when it comes up, without stopping a record from being complete.
At each exchange, by email or on the site chat, the agent files in the sheet what the person said, column by column, in their own words. What they did not say stays empty: the software rejects "not provided" or "unknown". Before writing, it rereads the conversation so far, so it does not ask again for what was already given. If a detail is missing, it can follow up on the days you set, as the page on the prospect who has not given everything explains.
It is not the AI that decides everything is in: it is the software, when every expected question has its answer. You then receive a card in "To do" and an email: the prospect's name and their request as the title, their answers to each of your questions, in your order, then who they are in a sentence or two, drawn from what they wrote. After that, the agent no longer replies to them alone by email: a new message from them waits for you, unless it answers a question the agent had just asked.
On the card announcing a complete record, the next step is offered to you: a meeting or a polite refusal, and, if you chose it at creation, the planned follow-on (a quote, a price list, a call back). The first is already written, the others are written when you pick them; you read and send in one click. To read this record before you commit, see how to tell if a prospect is serious.
A good set of questions is short. Start from your proposal, not from a template.
The first four make up the BANT method (budget, authority, need, timeline), a starting point; the last line often matters more.
To build your own set, put each question through three tests:
When an answer is missing, leave the cell empty. "Not provided" later reads as an answer, and nobody comes back to fill it in. To ask these questions in a first reply, see how to answer a quote request by email.
Autonomous does not mean unchecked. You set the level of autonomy of every agent, its limits cannot be crossed, everything it does is recorded in a plain-language log, and one gesture stops it all.
For the tasks you don't have time for. You give the address of your website, then describe the work in your own words. Nothing to install.