'Smart' auto-complete

By Emin Tatosian

From self-driving cars to automated image diagnosis, Artificial Intelligence is augmenting human activity in an increasingly meaningful way.

But you don’t need to embark on a multi-million-dollar project for your products to be usefully ‘smart’ too.

Take the humble auto-complete as an example – it helpfully filters a long list of items as you type so you can find an item quickly.

But it can be better; let’s explore how with an example.

Jack is a relationship manager at a global investment bank based in New York City, and covers clients in North America exclusively.

After every noteworthy interaction with a client, Jack enters a short note into the Customer Relationship Management system, which he does several times a day.

Jack’s just had a productive call with Barclays Capital Canada Inc and wants to enter a quick note.

Here’s how we can improve the client auto-complete to save Jack a few precious moments every day.

Alphabetically sorted

As Jack types Barclays, the auto-complete filters the list down to 125 from several thousand. But Jack’s client is buried deep down at 54th.


Location aware

Adding location promotes North American clients for quicker access.


Learns from usage

Clicking the auto-complete shows Jack’s top 5 clients based on the number of notes entered for each.

These small and relatively simple changes don’t transform the auto-complete into a self-driving car, but they do speed up a mundane task and save Jack a few precious moments.