The five automations worth building first

Most automation efforts begin in the wrong place: the messiest, most judgement-heavy process in the business, usually chosen because it causes the most pain. It also has the most exceptions, so the automation collapses under them and everyone concludes the technology does not work.

The better starting points are boring. They are the tasks a new employee could learn in a morning from a written note.

1. Answering enquiries out of hours

Enquiries arrive in the evening and get answered the next morning, by which time a competitor has replied. An assistant that acknowledges the message, answers what it can from your own price lists and policies, and flags the rest for a person, changes the response time from fourteen hours to two minutes.

The important part is the boundary. It answers from your documents or it escalates. It does not improvise on price, delivery dates or anything it would be embarrassing to get wrong.

2. Following up on quotes

Almost every business loses deals to silence rather than rejection. Follow-up is the first thing to disappear in a busy week, and busy weeks are exactly when the most quotes are out.

An automation that checks whether the customer has replied, drafts a short nudge referring to the actual quote, and stops the moment they respond, is unglamorous and reliably profitable.

3. Getting documents into records

Invoices, purchase orders and delivery notes arrive as email attachments and photographs, and somebody types them into a sheet in the evening. Extraction handles this well now, including photographed pages, and it can check what it read against the original order rather than just copying it.

Set it to flag mismatches instead of silently accepting them, and the finance team gets an exception list rather than a transcription job.

The goal is not to remove the person. It is to change their job from typing everything to checking the handful of things that look wrong.

4. The morning summary

Owners open five dashboards each morning, or more often, stop opening them. A short message before the working day, pulling yesterday’s orders, enquiries and payments into plain language with a comparison against last week, gets read.

It is also the easiest automation to build and the one that most reliably survives, because someone notices immediately if it stops arriving.

5. Handovers between people

A sale closes and three people need to know. A delivery fails and someone should call the customer. These handovers are where work falls through the gap, and they are usually held together by somebody remembering.

Automating a handover is simple: watch for the trigger, assemble the context, notify the right person, and record that it happened. It rarely saves dramatic amounts of time, but it prevents the failures that cost customers.

The rules worth keeping

• Automate what is written down; document the process before you build it
• Anything expensive to reverse waits for a person to approve
• Log every run in plain language, so you can see what it decided and why
• Start narrow, widen once you trust it, and keep a way to switch it off
• Measure the time it actually saves after a month, not the time you hoped for

What to expect

A first automation usually takes a few weeks, most of it spent understanding the process rather than building. The build is rarely the hard part. Agreeing what should happen in the awkward cases is.

That is also why the first one is the most valuable: once a process is properly written down, the next three get considerably faster.