Skip to content

Automations

See what it will do before it does it

An automation you cannot inspect is a thing you hope is working. Before anything is switched on, Leadmend writes out every step in plain English and marks it always, sometimes or never — and when the answer is never, it says why.

No account, no email, nothing saved. This is the real preview, running here in your browser.

What this business has set up

Change these and watch the routines below answer differently. Two are locked because they are not a choice an owner gets to make.

Opening hours set

Without them, anything conditional on being closed can never be true.

Alerts set up

An email address or Telegram chat for the AI to reach you on.

Email sending

On for every account. This is ours to configure, not yours.

Always on

Every routine below can run, and every step in them can reach somebody.

  • Chase an enquiry that went quiet

    When a new enquiry arrives.

    One click
    1. Alwayswait a week
    2. Sometimesemail the customer, written by the AIonly when that enquiry left an email address and hasn't unsubscribed — the AI writes it from what they asked, and your saved wording is sent if it can't
  • Ask for a review when the job is done

    When you mark a job done.

    One click
    1. Sometimesemail the customeronly when that enquiry left an email address and hasn't unsubscribed
  • Flag an urgent enquiry for a personal call-back

    When a new enquiry arrives and it looks urgent.

    One click
    1. Sometimestag them "urgent"only on triggers that have an enquiry attached
    2. Alwaysalert you
  • Follow up in the morning on anything that came in overnight

    When a new enquiry arrives and it's outside your hours.

    One click
    1. Alwayswait a day
    2. Sometimesemail the customeronly when that enquiry left an email address and hasn't unsubscribed
  • Come back round to a past customer months later

    When you mark a job done.

    One click
    1. Sometimesemail them again in 180 daysonly when that enquiry left an email address and hasn't unsubscribed by then
  • Answer an enquiry that arrives after you have closed

    When a new enquiry arrives and it's outside your hours.

    You’d build this

    You can build this today. Turn off “Opening hours set” above and the preview will tell you it can never fire — that is the answer we would rather you got here than after a quiet week.

    1. Sometimesemail the customeronly when that enquiry left an email address and hasn't unsubscribed
    2. Alwaysalert you

Right now this shows 0 routines that cannot fire and 0 steps that cannot reach anybody. An owner sees this before publishing, not after wondering why nothing happened.

Why the boring answer is the useful one

“Never” is a real answer, and it is on screen

Most automation tools have exactly one state: on. A routine that quietly cannot fire — because nobody set the opening hours it waits for, or because the channel it sends through was never connected — looks identical to one that works. It stays that way until an owner asks why nothing happened, which is usually the month they cancel.

Switch off one setting and our own demo breaks, on our marketing page

Turn off “Opening hours set” and the routine below goes from working to never — because a condition about being closed cannot be true when nothing knows when you close. We could have locked that switch on and nobody would have known. It is unlocked, because a preview that flatters the product is not a preview.

This is the same code, not a mock-up of it

The routines below are the library a customer installs from, run through the same sanitizer that guards the publish button, rendered by the same function that draws the dashboard. If we changed what a routine did and forgot this page, the page would change too.

What happens after it runs

The preview is half of it. Every run is recorded — what fired, what was skipped, and why — so an automation that ran invisibly stops being indistinguishable from one that never ran. Your dashboard shows the last seven days, and a weekly email brings it to you: what the AI handled, what was marked won, what is waiting on you, and one thing worth fixing.

The other two claims on this site carry the same kind of proof — the reply engine’s audit is published pass or fail, and the website builder runs for anyone without an account.

The product

Where a routine's work ends up

Every enquiry a routine touches lands here, with the reply that went out and the reason it was flagged.

The dashboard's enquiry list. A customer asks about annual boiler servicing and a monthly payment plan; the entry is marked Needs review at 75% confidence, with the line “Flagged for you — asked about monthly payment plan terms — needs a person to confirm what’s available”, followed by the reply the AI sent and a Sent to the customer marker.
Every enquiry, the reply that went out, and why the AI flagged this one for a person. Sample enquiries on a real account — the replies are the production engine’s.
An enquiry asking roughly what a heat pump costs installed. It is marked Needs review at 62% confidence and flagged with “asked for an installed price, which needs a person to quote”. The reply below it explains what the price depends on, asks a qualifying question and offers a site visit, without naming a figure.
Asked for a price, the engine refuses to invent one — it qualifies the job and offers a visit instead. The hard rule that produced this is published on /reply-quality.