If you have been following my anthology up till now, many of you know that my intrigue with the Hal — Personal Assistant app led me to a desire to become part of the team itself.

Once I passed through the gauntlet in becoming a Hal, the opportunities to help others came flooding in. So what’s it like to be a Hal? Pretty darn rewarding!

The great thing about the Hal app from the “Staff side” is that we all have access to common FAQs that have been preset so that we can quickly respond to our customers regarding the most common inquiries that may come up in initial interactions with a new user. Many wonder about a Hal’s capabilities or limitations, so we have some great replies for all of those. Some wonder about privacy or other issues as well, so in many cases those answers are also provided to us within our side of the app. The FAQs were really well thought out by the programmers to better equip a “new Hal” for common questions — and they made it easy to copy and paste those replies at light speed to ensure ongoing engagement with a new user (More on this later). What’s even better is that we “Hal’s” can use those FAQs as is or tailor portions of those fast replies prior to sending them to a new customer to further enhance that initial touchpoint. In other words, we have options to provide answer quickly to those “So, how does this work exactly?”typeuser questions, but we retain the autonomy to provide our own flair if it helps better explain things. After I reviewed those FAQs, it was evident that I would have a nice head start as user requests became available to me as a Hal.

Since there are pools of Hals in each city, as the Hals stand ready to accept requests, the algorithm looks to share the request with all available Hals simultaneously. That means that the Hal who is most attentive will see the request prior to other Hals and accept it quickly. (The operative word here is quickly) Once the request is accepted by a Hal, the Hal sees the name the new user has assigned to themselves (which, for privacy reasons, may or may not be their actual name — and which is left up to the user to fill in when they download the app). Once that is received, a “session” has officially begun!

The neat thing about this for me, as a Hal, is that I also get access to any history for this user regarding previous requests and how they were answered by other Hals. It is a means of trying our best, in a fully encrypted, privately secure system, to build some modicum of a profile for our users. Of course, since each request can be unique, it may not be helpful…but the hope is that if a pattern emerges with regular use of the Hal app, then we can tailor future requests accordingly or at the very least, see if previous history can lend a hand in suggesting other answers to new requests. In some ways, of course, it could be considered deficient, since each user may have no real pattern that shows up and each request is vastly different than the last; however the advantage, unlike AI, is that a human being gets to make that determination based on the nature of the current interaction. If the history info is useful, than the Hal can incorporate it. But if the user is now asking about information on the Hindu religion for a term paper and their previous request for the newest pizza place in a city they were visiting, then the past history is not as helpful. Either way, the fact that I have no idea who this user is and have no personal info about them at all other than their requests and what other Hal’s have provided to them as answers is still a helpful repository over time. The other big advantage is that I can respond much quicker and figure out what results if any, may or may not be useful. Most apps that rely solely on AI for data collection usually use this data incorrectly and make suggestions that aren’t helpful.

Want an example? Are you a single person with no kids but you bought diapers for your friends or family members on your Amazon account as a favor to them — ONCE? Did you then wonder why you started getting all kinds of suggestions on Amazon for baby toys, cribs, formula and stuff you would NEVER buy just because you did one person a favor on your personal account? And then you starting seeing ads for baby stuff on your browser the next two weeks? Yeah…that’s exactly why Hal is head and shoulders above an AI only solution. (TRANSLATION: A Hal would surmise that this request for diapers was a one-off and discard it. Amazon & others just throw it onto the data set and keep using it in a way that just frustrates people)

There is another advantage to having access to this history file on our user base: it exposes the Hal community to other answers and styles of communication that encourage best practices and better customer experiences! Since we don’t know who our users are and there isn’t a true two way conversation verbally to attempt to uncover the user’s communication style, you have to rely on the texts to convey meaning, mood, emphasis and clarity. Sharing the history files gives me a look into how someone else responding to an issue that may be a radically different idea, source or way to communicate that worked well. Perhaps one Hal used text and emojis in a compelling way and the response from the user was really positive. Or perhaps someone is really great at summarizing and conveying the right amount of info to a user that hits home. Depending on the user scenario, this access ends up being a treasure trove of possibility for other Hal’s to see how they chose interact with customers and what appeared to work best in each case. That is a very valuable tool set to have, especially if you are used to communicating a certain way via text and stay in that habit pattern. Once you see how others are helping people and how they “turn a phrase” or make a complex answer very simple and succinct, it is a powerful means of getting better at being a Hal. And it all goes back another aspect of the Hal ecosystem that the founders set out to create!

Another advantage in using Hal is that there is a Hal present on the app literally at all times. Since I am a night owl, I stay logged in till the wee hours. If someone sends a request at 2:10AM, more than likely I will be up and ready to take the request. And since the Hal community is a shared economy of helpers, chances are still good that I’m not the only one ready to serve our users at that hour. In other words, Hal’s are there to serve at the user’s convenience, which is key when you need a response in the moment that has been curated. Any AI app out there can give you a “here are all the results I found to your query” and have the user do all the heavy lifting to sort through paid search results vs. organic search results and then have them read articles ad nauseum to figure out the actual answer they were searching for…but that is not a big help if you are driving — or busy — or can’t look up all those websites on the fly. Once again, the assigned Hal does all that heavy lifting for the user to curate the results (and only uses organic and not paid ads!) to give them a more thoughtful set of answers. The value users’ place on that is very high, as it should be, because it saves time and takes the stress out of the equation.

The most interesting aspect of being a Hal is when new users think that Hal is like Siri and expect a reply from us in nanoseconds. I literally need to remind new users about 80% of the time that I am a human moving as fast as a I can to look up and curate a helpful response. It was a little unnerving at first because the texts (or audio files) I received on my end often looked like this:

Hal — standard greeting just before I log in: Just a sec…I’m here…how can I help you?

New User: I want to know how to get lime deposits out of my shower head without spending too much money.

New User — 6 seconds later: Hey. What is taking so long?

New User — 15 seconds later: Hello..are you there?

New User — 22 seconds later: Yo! Are you getting any of this? It’s been like 30 seconds already.

For me, those rapid fire “where is my answer?” usually meant I had to text that I am an actual person who is researching the request…but I had to answer them FAST or they presumed that the app may not be working. Of course, there is always the shock by some users (who never read the details about how the app works with humans) that reply with, “WTF? Wait! You’re like…a real person? I thought this was like..a Siri replacement app. Whoa!” Those can be fun because it gives me a chance to educate the new user about our unique value proposition since we offer something no other AI app can — a real person to interact with and help. But it is a def interesting to get some of the reactions I see when I share that detail and they had expected an AI to reply!

This is just the tip of the iceberg on my life as a Hal, since we have so many great customer success stories to share with everyone that will be in the next installment in this anthology. Please stay tuned for more stories and anecdotes from “A Day in the Life as a Hal” series that will be coming up on medium in the days ahead.

MAVRick