Big Plum Digital
Home
Products Contact Request a diagnosis Digital audit Plum Notes
ESENDA

Plum NotesAutomation

Automating with AI without losing the human touch.

AI can save you hours of repetitive work every week. It can also make your customers feel they're talking to a machine that doesn't listen. The difference lies in what you automate and what you leave in a person's hands.

Published
Reading time
3 minutes
By
The Big Plum team

In short

  • Start with a process that is repetitive, frequent and has clear steps.
  • Automate what moves data; leave to a person what needs judgement.
  • Anything that affects a customer or sensitive data goes through human review.
  • Measure the time you save before automating the next thing.
In this note
  1. 01The temptation to automate everything
  2. 02What is worth automating first
  3. 03Where AI really helps
  4. 04Where we wouldn't leave it alone
  5. 05A first project in four steps
  6. 06What to look after
  7. 07The human touch, closer

01The temptation to automate everything

With today's AI tools, almost any task looks automatable. And it is, halfway. The problem isn't that AI can't do things: it's that it does them with great confidence even when it's wrong.

That's why we don't start by asking what can be automated, but what is worth automating and what happens if it goes wrong.

02What is worth automating first

The best candidates meet four conditions: they happen a lot, always follow the same steps, take up time, and mistakes are easy to spot. Some examples we see often:

  • Moving contacts from your website into your CRM or spreadsheet and alerting whoever has to reply.
  • Sending appointment or payment reminders.
  • Sorting incoming emails and sending each one to the right person.
  • Preparing a draft quote from what the customer has filled in on a form.
  • Summarising meetings or calls and noting down the tasks that come out of them.

None of them replaces anyone. They all remove work nobody misses.

03Where AI really helps

AI shines with free text: reading, understanding, summarising and writing. It's good for:

  • understanding what a customer is asking for in a long email and categorising it;
  • pulling data out of documents, invoices or forms;
  • preparing a first draft reply for a person to review and send;
  • summarising a lot of information in a few lines.

To move data from one place to another, on the other hand, you often don't need AI: an ordinary automation is enough, and it's cheaper and never improvises.

04Where we wouldn't leave it alone

Some decisions need judgement, context or empathy, and there AI suggests, but a person decides:

  • the final reply to a complaint or an angry customer;
  • prices, terms, contracts and any commitment;
  • sensitive, health or personal data;
  • decisions about people: hiring, rejecting, disciplining.
AI can write the draft. The signature is still yours.

In our projects we put it in writing from the start: what goes through human review and what doesn't.

05A first project in four steps

  1. Describe the process as it is today. Who does what, with which tools and how long it takes. Without idealising it.
  2. Pick the step that eats the most time. Just one. The first project has to be small so it goes well.
  3. Automate it with a person at the end. Let the machine prepare and a person review, at least at first.
  4. Measure for a month and decide. How much time it saves, how many mistakes it makes, what the team thinks. With that you'll know whether to expand, adjust or drop it.

06What to look after

  • Your data. Which tools see your customers' information, where it's stored and under what contract. Not everything is fit for everything.
  • What happens when it fails. Every automation fails at some point. It has to alert someone and have a plan B, even if that's doing it by hand.
  • Don't let it depend on one person. Document what each automation does and who maintains it.

07The human touch, closer

Done well, automation doesn't push your customers away: it gives you time for the conversations that matter. Less copying and pasting, more listening.

If you want to see which part of your work could be automated, in AI automation we explain how we approach it, or you can start with the free diagnosis.

Keep reading

All notes