Teach My Little Sister How To Drive is an upcoming DMV simulator built around an unusual driving lesson. From the passenger seat, players guide a chaotic young learner through her test while communicating with her through a live AI chatbot.
That central idea has created a major financial challenge for Easy Fox. The developer says the free demo’s rapid growth has pushed AI operating costs beyond $1,000 per day, forcing the four-person team to take out a bank loan to keep it available.
Player growth has multiplied the running costs
In a recent update, Easy Fox said the demo audience had increased more than twentyfold over the previous month. The team thanked players for the unexpected attention, while warning that the extra activity had considerably increased its operating expenses in the Steam page’s latest update.
The demo initially connected players with Google’s Gemini. With no usage charge stopping curious players from experimenting, the chatbot began returning limit errors. Easy Fox says Google provided no simple remedy, so the developers switched to an OpenAI model while still hoping to resolve the Gemini issue.
A free demo costing over $1,000 every day
Easy Fox says maintaining the demo costs more than $1,000 each day. That expense may ultimately need to be reflected in the price of the completed game. The team would like to keep the demo running, but says the financial pressure could force it to close before the full release. No final decision had been made when the update was published.
The developers also addressed complaints that the AI-controlled sister does not reliably follow player instructions. Unpredictability is part of the game’s premise, but inconsistent responses add another complication to a project already dependent on expensive, continuous AI processing.
An AI agent at the centre of play
AI is increasingly used during game production for tasks such as coding and asset creation. Easy Fox’s approach is different: an AI agent remains active during play and drives the conversation that forms the game’s main interaction. The concept recalls earlier voice- and conversation-led games such as Lifeline and Hey You, Pikachu!, although modern language-model usage brings an ongoing per-interaction cost.
That makes the project an interesting contrast with the hardware featured in RETROCADE’s racing simulators. A physical driving cabinet has its own costs, but does not require a cloud AI service to generate every exchange. Readers comparing interactive entertainment setups can also use RETROCADE’s arcade machine buying guide or browse the broader arcade news and gaming coverage.
Popularity can become a liability
The situation illustrates a difficult business problem for AI-powered games: a successful free demo can increase costs without creating matching revenue. Every player interaction consumes computing resources, and sustained real-time use can be far more expensive than employing AI only during development.
Similar cost pressures have affected AI-generated video, commercial artwork and other services, with providers introducing usage limits and more expensive plans. The wider debate has been described as a “tokenpocalypse”, while examples such as horrifying fast food posters show how quickly the technology is spreading into commercial work.
For Easy Fox, the issue is immediate rather than theoretical. A popular demo has created a daily bill large enough to require borrowing, and the studio may need to balance the final game’s pricing and design against the cost of keeping its AI-driven passenger online.
Leave a Reply