Yuriy Dzyuban, a backend developer at IT company Master of Code Global, created a virtual AI traveler who moves between real landmarks and takes selfies against their backdrop. Now Teddy (Theo) is conquering social media and becoming a travel influencer.
Yuriy Dzyuban, a backend developer at IT company Master of Code Global, created a virtual AI traveler who moves between real landmarks and takes selfies against their backdrop. Now Teddy (Theo) is conquering social media and becoming a travel influencer.
dev.ua talked to the developer and found out what makes such a traveler useful and how it was created.
The idea to create a «traveling toy» came to Yuriy sometime in 2017-2018, when he was learning to program. «One of my self-study projects was about creating a kind of social experiment: to see how far a real toy could travel around the world, passing from hand to hand, including between strangers,» he says.
That project, according to the developer, never reached launch, but Yuri still has the toy with the QR code.
«But 7 years have passed, ChatGPT, Midjourney, Sora and other models have appeared, and at some point I thought, why not create a slightly different (although perhaps somewhat similar) project? Namely, a virtual traveler who moves between real landmarks and takes selfies against their background. It seems that in modern terminology this can be called a travelling AI-influencer,» says the developer.
He adds: if the current project with an AI influencer finds its audience, there are ideas on how the old project could be «woven» into the current one.
The creation of the project, says Yuriy, took about 30-40 hours during May 2025. «Creating a project means formulating an idea, selecting, detailing and visualizing the character, developing the actual content creation algorithm, and, finally, preparing content for a trip to the first country (Ukraine). In fact, creating this project is in many ways similar to writing a work of fiction in the chat fiction format, with quite active involvement of various modern AI tools,» says the developer.
According to him, creating content for one day of travel (4-10 Twitter messages plus 2-5 images) takes 30-60 minutes. A few more hours can be spent preparing the route and «adventures» around the country as a whole.
«Except for my own free time, the project cost $0 to create,» he says. The developer uses a number of AI chatbots such as ChatGPT (mostly), Gemini, Claude, Grok — all in free versions. Yuriy also uses the free Perplexity plan quite actively (ideation at various stages of route planning). Images are generated in OpenAI Sora.
«Since I usually only have about an hour on weekdays for pet projects (for example, in this case, in this time you can create content for 1-2 days of a virtual trip), the daily limit of 12 images on the free Sora plan is usually enough for a day,» he says.
Thus, if desired, the IT expert says, this project can be created entirely in the free version of ChatGPT. «Other chatbots turned out to be less suitable for writing Twitter messages in the style I needed (somewhat unexpectedly, Grok also did not impress),» Yuriy dreams.
As of now, https://x.com/SeenByTeddy is an experiment/MVP that one person (namely Yuriy) is working on in his free time from work (1-2 hours on weekdays and a few hours on weekends).
As Yuriy already noted, this project is essentially an example of chat fiction, that is, a work of art in the format of messenger messages, about travel, adventures and places of interest. «That is, the focus of the project is entertainment and educational, and the potential audience is seen as fans of fiction (for some reason, I primarily see middle-aged women with children), travelers, and simply people interested in the capabilities of modern AI. However, these are my assumptions, and the real goal of this project can be called testing the hypothesis of whether such an AI influencer could be interesting to someone,» the developer explains.
There are also purely subjective motives — like playing with Sora, understanding the possibilities of automating such tasks, and just for fun and showing children an example of what can be created.
Yuriy explained how he used AI services to create Theo. According to him, the following main types of work can be distinguished in this project, in which the following technologies and tools were used:
«Currently, the project is being implemented mostly manually, with a manual workload of around >60%. All requests are made through the web interface of the relevant tools (not through the API). I assume that automation can be increased to ~¾, but ultimately things like humor and emotional nuances are currently (or in principle) poorly automated, and are unlikely to be possible without human moderation,» he says.
Even before the first Twitter message was created (June 2, 2025), Yuriy generated all the content for the AI bear’s journey through the first country, Ukraine. These are 105 Twitter messages and 52 images, which will be published over 16 days. During this time, Theo will visit Kyiv, Chernobyl, Lviv, the Carpathians, Chernivtsi, Odessa, Zaporizhia, Kharkiv, and a number of other places.
«At this stage, the goal is to validate the idea itself and check whether such a project will be interesting to anyone. Specifically, the task now is to overcome the problem of the „chicken-egg“ cycle, that is, to get in the eyes of the audience and get feedback from them. To do this, the influencer will visit Ukraine and, possibly, several neighboring countries,» says Yuriy.
If the idea «takes off», then the developer’s plans include:
Yuriy also has two ideas for monetization, but he is in no hurry to talk about them. «That would probably be presumptuous right now,» he says.
If the project doesn’t take off, the developer won’t be upset. «Well, that means I’m one step closer to another, the same project,» he philosophically sums up.
Recall that previously Yuriy created a book for his children using ChatGPT and Midjourney in just one day.
