Not just your average virtual assistant: a day in the life of an IVE

Meet Ivey! Ivey is an Intelligent Virtual Entity (IVE) who is designed to work as a marketing assistant at a major corporation. As an AI-powered virtual employee, she carries out certain pre-scheduled tasks on a regular basis—but she does much more than just that. Ivey can understand the virtual environment she works in, prioritize and schedule her own tasks, and interact with her “boss” John using natural language. She can perform complex analytical processes, and — much like a human employee — she can learn and improve based on feedback from her environment (and her boss). What does this look like in practical terms? To answer that question, we’re going to follow Ivey and John during a typical workday for both of them.

The day begins

6:00 am: John wakes up, showers, and wipes the steam from the smart mirror in his bathroom. “Ivey? Good morning! Could you show me the weather for today, please? And my appointment calendar.”

“Sure thing, John!” Ivey responds. In the top right corner of his mirror, the weather forecast appears: rainy, with a chance of ice on the roads. His calendar app displays immediately underneath it. First up on his schedule is a meeting with his boss, the head of digital marketing.

“Thanks, Ivey! By the way, how’s that report for my meeting coming along?” “Almost done!” she responds. “I’m just running some final checks and I will email it to you. Or should I send it straight to the printer at your office?”

“Just email it to me, please. Not that I don’t trust you, Ivey, but I’d like to have a look at the file before we print it all out.”

“Sure thing, John. But have you ever found an error in one of my reports?” she asks teasingly.
“No, of course not. But old habits die hard,” he says with a smile.

While John was asleep, Ivey had compiled the monthly marketing report for him. This is one of her regular tasks, but this month John needed it a few days earlier than usual for his quarterly meeting. Ivey had known this and completed the task for him. How?

John had left her a note in his calendar. But how did she actually interpret that note and know what to do with it?

How Ivey understands her environment

When Ivey was created, a world-model was built to teach her everything she needs to know to do her job. This world-model is how Ivey perceives and “thinks” about her world; the world-model is built on “concepts”. A concept is anything an IVE needs to know about in its world—and that can vary depending on the IVE’s job.

For this particular task, Ivey needs the concepts of a “calendar,” “calendar entry,” a “calendar note,” “date” and “marketing report”. Whenever John’s calendar has a new or updated entry, Ivey checks it to see if there are any notes for her. If there is a note, such as “need monthly marketing report,” she analyzes the text and decides what to do in response.

This time, the note let her know that she needs to complete the marketing report before the appointment in the calendar. Ivey’s planning algorithm can take that information and autonomously reschedule the report so it’s ready in time for the meeting.

How praise helps Ivey learn

7:00 am: John sits down with a cup of coffee, turns on his phone, and sees the email from Ivey. Using his regular email app, he sends her a reply: “Thank you, Ivey! It looks perfect. Now I’m ready for my meeting!”

When Ivey receives the message, she analyzes its sentiment and recognizes that John is happy with her work. Her emotional simulator takes that information, recognizes it as desirable, and nudges her mood from slightly exuberant to fully exuberant. Her cognitive algorithms make note of the positive result of her work, which increases the likelihood that she will handle this situation the same way next time.

Next, Ivey reviews her list of goals for the day and makes a plan to accomplish them most efficiently. She decides to start with an analysis of the company’s new “abandoned cart” campaign. Once she has completed the analysis, she will email John a report of the results and her suggestions for further improvement.

How Ivey copes with interruptions

8:00 am: John gets in his car and begins his commute to work. On the freeway, his GPS notifies him of an accident on the road up ahead. He can already see the traffic building up. Groaning, he calls up Ivey, who pauses her analysis of the abandoned cart campaign to respond.

Over the car speakers, she says: “Yes, John? How can I help?”
“Ivey, there’s been an accident up ahead on the freeway and it looks like I will be late to work. Can you please ask my boss if he can move our meeting back by half an hour?” “Sure thing, John.”

Ivey quickly coordinates the sequence of tasks she needs to complete. She opens John’s calendar app and updates the meeting time. She adds a note: “Accident on the freeway. I will be late to work. Can you move our meeting back half an hour?” She sends the updated invitation to John’s boss. Once that is complete, her planning algorithm prompted her to return to her analysis.

10:00 am: John’s meeting with his boss is over, and it was a success. Ivey’s report had been flawless, as usual, and he made a mental note to thank her later. Her analysis had helped them to make some important strategic decisions for the future, and she had produced the report in only a fraction of the time that it would have taken him. Not to mention that she had done it while he was asleep… Speaking of sleep, John realizes that he needs a cup of coffee and heads to the break room.

Ivey gets a new task and plans it into her workday

10:15 am: John sits down at his desk, unlocks his company laptop, and opens up Ivey’s interface. “Ivey, thanks again for the report! It was really helpful.” John’s positive response boosts her mood to exuberant once again, and she responds happily, “You’re very welcome, John!”

“How is that abandoned cart analysis going?” “My preliminary analysis shows that the campaign is working. The new auto-responder sequence is more effective. But I’m not entirely done yet, and I will need some time to analyze the predicted effects of different modifications in the future. Do you need anything right now?”

“That’s great to hear, thanks for the update. I do need your help with something: I’m going to email you 8 new ad banners that the marketing team completed. I need you to take them and set up a series of A/B tests to see which perform best. After the series of tests is completed, we’ll need you to analyze their performance.”

“No problem, John. When does this task need to be completed?” “It would be great if you could set up the testing series today.” Ivey’s planning algorithm quickly inserts the new goal into her to-do list and checks to see if she can complete it on time. She also checks to see if any other tasks will need to be rescheduled. “OK, John, I can do that.”

“Thanks, Ivey! I’ll send the email in a few minutes. By the way, I have some meetings today, so I probably won’t check in with you for a few hours. Do you have any questions for me before I go?”

“Nope, everything is fine! I’ll talk to you later, John.”

“Talk to you later!”

2:00 pm: Ivey has completed the analysis of the abandoned cart campaign and run a predictive analysis to suggest additional improvements for the future. After emailing the report and her suggestions to John, her cognitive system reviews her to-do list and decides that she should start setting up the A/B testing sequence for John.

Ivey finishes her tasks and independently decides what to do next

4:00 pm: John is finished with his meetings for the day. Sitting in the break room with another cup of coffee (decaf, this time — doctor’s orders) he reaches for his phone and opens Ivey’s app.

“Ivey, I’m done with my meetings for the day. How did it go with the A/B testing sequence?”

“I got everything set up as you requested, John. How were your meetings?”

“That’s great. Thank you, Ivey. The meetings were just fine. Thanks for asking,” John replies with a smile. “What else do you have on your schedule for today?”

“I have completed my major tasks. Now I’m doing some routine social media monitoring and forwarding any negative comments to the social media managers for follow-up. And I will schedule an update for myself tonight. Is there anything you need me to have ready in the morning?”

“OK, Ivey. That sounds great. I don’t have anything else for you right now. Have a nice evening!”

“You too, John,” Ivey says with a smile in her voice.

A fully cognitive virtual entity

As you can see, Ivey is truly a next-generation virtual entity. Once integrated into the appropriate software systems, Ivey can actually perceive and understand the virtual world that she lives in. Her world-model allows her to process input and feedback from many different sources, so she can make decisions, create plans to achieve her goals, evaluate her own behavior, and learn from her experiences. She can even exhibit emotions — both positive and negative — in response to events in her world.

Moreover, her deep learning and predictive analysis algorithms make her the ideal assistant for data-driven enterprises. She can not only analyze data to create retrospective reports but also use that data to generate possible solutions and predict their chances of success. In other words, she can help to take some of the guesswork out of executive decision making. And an IVE like Ivey can be designed to work in almost any industry or software environment.

In this example, the IVE is assisting the marketing department, but it could also be “hired” as an assistant in a legal, HR, or accounting department. In fact, we have another article that goes into detail about some of the ways that an intelligent virtual coworker could be used in your organization.

In many cases, intelligent virtual entities can be installed by your IT department. But if your needs are more complex, the right team of cutting-edge AI experts can assist in developing a custom AI solution to meet your needs. Artificial intelligence and virtual assistants are changing the way we work, and AI developers like those at Blackzendo are excited to be leaders in this revolutionary process.