AI Personalization at first bite

AI Personalization at first bite

Yesterday I ordered a pizza with a bot. Yes, a Pizza bot!

It made me think back about the changes the pizza ordering process made in the past 30 years, and its part of a huge movement in Pizza but more importantly in the discovery stage (search and advertising), our customer loyalty as well as what it could mean for privacy.

It all began at first bite.

Pizza ordering throughout time

At the age of five (30 years ago, give or take), when our family desired a pizza pie (I pushed for olives), we would leave the house to a local one or make a tad longer trip to my uncle’s Pizzeria. Yes, we had to leave the house.

At ten, our door and mailbox (snail mail) was already overflowing with fliers, and the heavy Yellow Pages book was available to scour additional options as well as call for a delivery. It would arrive 40-50 minutes later, and yes with green olives topping it.

During my early twenties, the Internet opened a whole new world of possibilities for pizza lovers (and beyond). We could search for pizza and get local or branded options, as well as loud, obnoxious (not at the time) and brightly colored indexes to select from. You make your choice, call them up, order a pie and have it delivered. My girlfriend at the time liked Feta, so we asked to sprinkle some over…you guessed it, olives.

Throughout time, our pizza choices has grown, but the results have been updated: due to regular economics as well as search engine parameters for relevancy, advertisement, timing and more. The better the attempt at answering to your need, the more likely you will click…and then make an order.

Even more recently, 7 years ago, a new option appeared which was searching for pizzas within an App Store. Results included pizza games, recipes as well as apps to order directly from within…no conversation required. With a few screens and clicks I could order my favorite olives topped pizza to be delivered to me.


The pizza bot experience

So as I mentioned, just yesterday I had my first pizza bot experience. You know, a conversation with an “artificially intelligent” interface which isn’t really ‘intelligent’, rather a more complex workflow; like the IVRs of yesterday or a simple and automated process.

The conversation went smooth, with besides asking what pizza I wanted, it inquired on the type of dough, whether to add a drink, where to deliver to, and payment method. It didn’t yet know of my love for Olives, or my wife’s recent birth and craving for Pepperoni, but still, it was right within FB Messenger, so I was able to remain in my current status, keep chatting with friends and make the order. Fun stuff right?

AI personalization experience

How will AI personalization evolve in the coming years?

My very own virtual personal assistant (VPA) will order Pizza for me.

My VPA will know olives is my favorite topping, and my dislike for Tuna, dough type and to order Saturday evenings, as I begin a diet every Sunday morning.

A possible scenario could be the following.

(By the way, I use Cortana as it is pre-installed on my work PC with all my details etc.)

Me: Cortana, would you order me pizza tonight?

Cortana: Sure, your regular with olives?

Me: Yes

Cortana: I noticed you’re meeting friends, so should I order a 2nd the same and a 3rd without gluten, like you ordered last time?

Me: Yes, thanks

Cortana: …and your wife isn’t home, so would you like beer instead of coke?

Me: you’ve read my mind. I mean, Yes!

Cortana: Prior to ordering, I found another local Pizzeria has 3 for the price of 2 and offer a no Gluten option. It saves you $5 total. Should I switch and order from here?

Me: You know what? Yes

Cortana: Ok, I will make that…

Me: Wait. On second thought, order from the regular place, I don’t want to take any chances.

Cortana: Ok. I’ve made the original order, total is $40 and is scheduled to arrive within 10 minutes of this evening’s game. Let me know by 8pm of any changes.

Times they are a-changing…

The above isn’t far-fetched, and it highlights 3 important areas that will be a focus towards ultimate personalization.


Today’s search engine provides multiple results (both organic and paid), which hopefully direct us to where we want to be; in our case, the right Pizzeria. We’re also aware of the revenues engines grow thanks to the usage and advertising, with the latter not always so distinguishable (or click avoidable) …especially in apps and games.

In the scenario above, there is only ONE result my VPA knows of, and possibly one additional promotion personalized for my needs.

This is because my VPA knows me better than I do and won’t recommend otherwise.

This change will decrease advertising traffic, yet will improve relevancy, as less potential impressions would mean higher costs, which would mean a greater attempt to answer my specific need.

There is no more ‘if you aren’t in the first page, you don’t exist’; it’s more…if you aren’t THE result, then you don’t exist.


In today’s digital world, loyalty isn’t etched in stone as in the past. We move from one service to another, product to product, and brand to brand. This is due to many reasons, including our personal experience: cost, relationship, language, honesty, transparency and more.

Well, my VPA knows me well and gets better with every action I take. It knows what my regular is, where and when to order it, how to pay including special occasions. It’s likely that one day it will make the regular order without my request at all.

The Pizzeria on the other hand would have to make sure its relationship with my VPA is awesome, otherwise I’ll be gone. In other words, it better treat it well and offer a few perks here and there.

So businesses will have to be extremely relevant, specific and personalized on a high-standard (product, timing and flexibility) to win loyal customers.


Combining discovery and loyalty brings up the area of privacy, or my personally identifiable digital footprint, which is exposed on the net to gain the benefits the data reveals. What impact will this footprint have on “Privacy Policies”?

When I will feel certain to trust my VPA, I will let go of my fears and provide access to my life. But what would happen if this data leaks? What if my VPA will be affected by a hack and the no-gluten pizza arrive regular, and an allergy attack may arise?

I believe privacy must become a core part of the backbone of my script to become a reality. As without impenetrable privacy, I likely won’t take part of the play to begin with.

AI Personalization, here I come

In 1-2 years, I will still begin a weekly diet on Sundays and Google will still make money from ads, but advertisers will need to put in more effort to be relevant and pay more to be seen; Pizzerias would need to adapt to VPAs; and the gatekeepers to our private information will understand what major responsibility they hold and its effect on personalization.

I do hope my VPA will care for me enough to know what’s best for me, and instead of ordering a pie Saturday evenings, it will sign me up to the gym.


Ariel Kedem
Ariel Kedem

Ariel Kedem is an entrepreneur, manager and foodie. He is the VP Products at Knowmail.

Ariel Kedem

Leave a Reply

Your Email address will not be published. Required fields are marked *

  1. Thank your for an informative, funny and thought provoking yet informal article. Good job (My apologies I teach – forgive the pedant conversant mode).

    Whenever the AI VPA comes into being your (Oh, our collective) privacy goes with its “coming”

  2. Pingback: The Pros and Cons of Chatbots - Knowmail

  3. Pingback: Should Artificial Intelligence Care for the Terminally Ill? - Knowmail

  4. Pingback: 7 Personal Chatbots For Your Workday - Knowmail

  5. Pingback: How can we trust virtual personal assistants? - Knowmail

  6. Pingback: An Introduction to Virtual Personal Assistants - Knowmail