LeadLoup
STANDBY
Book a callFR

Custom AI agents

By 8 a.m. the file is qualified, the quote is drafted, the follow-up is scheduled.

The agent reads every request, decides what to do with it under rules you wrote, then carries the file through to the follow-up. It reports back once a day, and the decisions that bind your company stay yours.

Book a call

28.2%

of Canadian businesses using AI use it for virtual agents or chatbots, the third most common application, behind data analytics and text analytics.

Statistics Canada, Canadian Survey on Business Conditions, Q2 2026

One day, hour by hour

Demo

Typical hourly profile for a service business. Demonstration values, not operating data.

Where a custom agent fits, and where it does not.

Yes, if

  • Requests come in at night or on the weekend, and the file does not move until Monday.
  • Days pass between a client's first question and the quote going out, not because the job is complex, but because nobody had the time.
  • Your follow-ups depend on somebody remembering to make them.

No, if

  • You get a handful of requests a month. Following up by hand still costs you less than installing anything.
  • You want an agent that passes for a person. Ours introduces itself as an agent in its first message.
  • Your work follows a fixed rule and calls for no judgment at all: that is process automation, not an agent.

What happens to a file, from first message to follow-up

  1. The signal comes in

    Text message, Messenger, email or web form: the agent answers in under two minutes. The delay is the same at night, on weekends and on holidays.

  2. It qualifies

    Service wanted, budget, timeline: it asks your questions, adjusting to what the client has already said. The answers land in your CRM with nobody retyping them.

  3. It prepares

    Client record, draft quote, prefilled form. You get a document to check and send, not a blank page to fill on a weeknight.

  4. It follows up

    A client who says they will think about it, or who goes quiet: the agent rereads the exchange, judges whether a follow-up is warranted, and writes to what was actually raised.

  5. It reports back

    One summary a day, at the hour you pick: what moved, what is waiting on a decision, what it could not handle. Nothing gets decided in silence.

Text message · after hoursDemo
  1. Hi, are you still quoting roofing jobs? I need something fairly quickly.
  2. Evening. Yes, we quote roofing. So I get this right: is it for a house or a commercial building?
  3. A house. Bungalow, in Laval.
  4. Got it. Do you have a deadline in mind, and is the roof leaking right now?
  5. A bit of water since last week. Sooner is better.
  6. File qualified · urgency flagged · record sent at 11:09 p.m.
  7. Visit offered for 8 a.m. the next day · confirmed by the client at 7:41 a.m.

What you get

11 deliverables

  1. The agent configuredOn your real channels: text, Messenger, email, web form.
  2. Your qualifying questionsWritten for your trade, not lifted from another client.
  3. The calendar connectedBooking without double-booking.
  4. The document templatesClient record, draft quote, prefilled form.
  5. The follow-up rulesWhen to follow up, what to bring back up, and when to stop.
  6. The CRM connectedOr a plain tracking sheet, ready from day one.
  7. The handover rulesCases outside the frame go to the right person at your end.
  8. The daily summaryWhat moved, what is waiting, what it could not handle.
  9. One sample per channelSo your team sees exactly what the client receives.
  10. A run on real casesBefore it goes live with your actual clients.
  11. An adjustment a few weeks inBased on the files that really came through.

What people ask us

How is this different from an automation?
An automation runs a fixed rule: if this, then that. It does exactly what was planned, and it breaks the moment the situation steps outside the frame. An agent reads the situation, decides what to do under your rules, and flags what it cannot handle instead of inventing an answer. If your work follows a fixed rule and needs no judgment, automation will cost you less.
Will a client notice they are not talking to a person?
The agent introduces itself as an agent in its first message. It does not pass for a human. Clients who want somebody from your team say so, and the agent hands the conversation over under the rules you set before launch.
What happens when a question comes in that nobody planned for?
It flags the question instead of inventing an answer. The conversation is handed over under your rules, and that question then widens what the agent knows for the next client who asks it, with nothing to reconfigure on your side.
How long before the agent is running?
Scope sets the timeline. One agent on a single channel, text for instance, goes live faster than a rollout across several channels at once. A real schedule comes out of the discovery call, once we have seen your channels and your message volume.
Do we keep control over what the agent says?
Yes. The answers follow rules you approve before launch, and you get a daily summary of every conversation. Nothing goes out before the frame has been set with you, and the rules can be tightened at any time.
Can one agent cover several channels at once, text and Facebook for example?
Yes, with the same qualifying rules behind each one. Conversations reach you in a single place no matter which channel the client wrote to first, and the daily summary groups them together.

See what this agent would do inside your business.

Book a call

20 minutes with David, founder of LeadLoup, no commitment. You leave with a concrete picture of what the agent would do on your real channels, and a clear timeline if a rollout makes sense for you.