Ansett
Om oss
Kunde
Our customer is a startup developing an AI-powered companion application that is intended to help lonely people get social activities and support regardless of time and location.
Detaljert informasjon om kunden kan ikke utleveres i henhold til bestemmelsene i NDA.
Utfordring
Developing an AI companion application requires fine-tuned neural networks that are able to interact with users in a human-like way. At the same time, the AI must not show any toxic behavior, spread hate speech, or give harmful advice.
In addition to that, to keep the application running, a certain business model should be developed to monetize the service and cover the development and maintenance costs.
Løsning
Kunstig intelligens
The core of the solution is the companion’s natural language processing engine. Each companion stands as a separate session covered by the custom-made artificial intelligence algorithm that dives deeply into the conversation between the user and the companion to understand the contest and act in the most human way possible.
Teamet vårt har også finjustert to algoritmer for datasyn. Den første sørger for nøyaktig posisjonering av ledsagerens 3D-modell i det fysiske miljøet ved hjelp av smarttelefonkameraer. Den andre modellen tar sikte på å gjenkjenne bildene som brukerne deler med ledsagerne sine, slik at konteksten for samtalen kan utvides og brukerne kan få mer hjelp.
During the conversations, companions learn about their users’ interests, hobbies, and personality traits. This allows them to hold a conversation as in real life, nearly wiping the border between physical and digital interactions. Also, this allows for tracking mood and situation changes in the background and giving care about the user in the form of questions, advice, and supportive messages.
3D-modeller
Our team has not only fine-tuned AI models to bring the most live experience on the market. In addition to that, Innowise Design Studio prepared a wide range of 3D assets that can be used both in mobile and AR modes.
The models include different alterations of the companion’s looks like hairstyles and clothing. This feature allows every user to create a personally appealing companion and increase engagement in interactions with it.
Also, while in mobile mode, a companion has its own space that can be decorated with furniture, plants, and other assets that can show the personality of both the companion and the user.
Inntektsgenerering
For å dekke kostnadene ved utvikling og vedlikehold har teamet vårt også implementert flere måter å tjene penger på applikasjonen.
The first one is introducing the in-app currency that users spend on assets and new datasets for their companions. With its help, they can turn a virtual friend into a music nerd, a tabletop games lover, a sports fan, etc. With the help of these datasets, users can make their companions even closer to themselves.
By purchasing new assets, users can alter the looks of their virtual friends and add some flavor to their digital spaces.
The second way of monetizing the application is the subscription. While all users can utilize the application free of charge, the subscribers get access to the most advanced AI models.
Personvern for brukerdata
To comply with the regulatory requirements and keep user data safe, we implemented an array of processes for anonymizing and encrypting data before it’s used in any way. Each user session is held separately but the application servers don’t track any personal data.
Prosess
Prior to developing the application, our team conducted an in-depth market analysis to come up with the target audience and applicable monetization models.
Our team suggested treating users who are 21-40 years old as the primary target audience. People at that age are usually working hard on building their careers and future and are not always able to find the support they need from the people around them.
After that, Innowise began development. We used the Scrum approach with one-week sprints and meetings three times a week to discuss important issues. Slack and Jira were used to manage tasks and discuss projects within the team. Each week, the project manager reached out to the client via Skype to provide an update on the development process.
Team
Expected results
As a result, Innowise developed a virtual companion application that allows users to create an AI-based digital friend and share their thoughts with someone who’s always around.
We expect the application to experience an increase in downloads and activation rates, along with rises in active subscriptions and in-app purchases.
- 2018 - 2022
10
300,000
