One morning I sent a new page on this site to a group of friends and asked them to try it.

The feedback came back the way feedback really comes back: phone screenshots on WhatsApp. A button that stayed grey. A picture that stretched into an odd shape. A form that would not move to the next step. Nobody wrote a report. Nobody filled in a feedback form.

I gave the screenshots to AI exactly as they were.

It read each image, worked out what each person was pointing at, and sorted the pile into two lists: three faults to fix, and a longer list of lessons about what a visitor expects. The three fixes were running on a test copy of the page the same morning. I looked at each one and decided what went live.

Feedback in your business already looks like this

A photo of a delivery order with one line circled. A screenshot of an error message. A forwarded chat from a customer. Somewhere in your company, someone retypes all of that into a list before anyone can act on it. Most of it never gets retyped, so most of it is lost.

AI changes the starting point. It can read the photo, the screenshot and the chat as they are, pull out what each one says, and give you a sorted list. The person who used to retype becomes the person who checks the list.

The deciding stays with a person

AI read and sorted that morning. I chose what went live. That split is worth keeping wherever you point AI at feedback: let it do the reading and the sorting, and keep a person on the choosing, because the choosing is judgement about your product and your customers.

Try it this week

Pick one channel where feedback already piles up: the customer WhatsApp group, the complaints inbox, the photos your outlets send in. Collect one week of it in one place. Give it to AI and ask for two lists: things that are broken, and things customers wish for. Read the lists with the person who owns that work.

The first time a morning of screenshots turns into a sorted list before lunch, you will know which other piles to point it at.

In your department

  • Operations & Delivery: Site photos and defect screenshots from your teams can go to AI as they arrive; it returns a sorted issue list for the morning meeting, and your supervisor decides what gets fixed first.
  • Sales & Marketing: Customer complaints and enquiries arrive as screenshots and forwarded chats; AI can group them into themes you can answer, and a person still writes the reply.

See where AI could start in your company. Enter your company name and see suggestions across people, process, technology and money. About two minutes. I read and answer every enquiry myself.

Find your starting point