Reality Check. You made a great game. Nobody found it.

Discovery is broken. And the chatbot era is accelerating the collapse.

Steam publishes nineteen thousand games a year. The average gamer can name fifty on a good day.

Younger players gave up on platform discovery years ago. They ask friends, influencers, and increasingly, an LLM.

Older players retreated to what they already know. Steam's own data shows two thirds of PC playtime goes to games more than six years old. The industry responded with over two hundred remakes since 2012. "Play something I already know" is now the rational choice.

Players are frustrated and creators are spiraling into a race to the bottom on advertising spend.

That is what a broken system looks like.

LLMs are making it worse, not better. Shorter shortlists. Heavier bias toward whatever dominated their training data. New releases and mid-to-long tail are squeezed out.

This is not a marketing problem. It is a foundational problem.

It cannot be solved by optimizing an app store. It cannot be solved by another AI bolted on top of the same broken signals.

Data on games is fragmented, shallow and unreliable. It is the infrastructure that must be fixed.

And the solution needs to bring control back to both creators and gamers.

We have established a single verified data source per game with a structure and governance model inspired by proven Internet structures. The pointer system provides tools for creators and publishers to generate standardized rich title descriptions.

We index the pointers, not the data, so creators retain full ownership and full control over how their titles are described.

Crucially, this architecture protects creator data from LLM harvesting, and rebuilds the foundation for how games are discovered, creating a new, open layer for game data that anyone can build on.

We offer human-centered discovery that represents your taste accurately, privately, and persistently. You express it in your own words, not by picking preset categories.

You get your own AI agents, rather than having one huge probabilistic one-model-fits-all applied to you. Architecturally distinct from LLM-first systems, it guarantees zero hallucinations, continuous learning, and complete data privacy.

We have a fully developed, running demo.

Source of Truth AB has established the data source on behalf of a non-profit in oversight of games industry organisations and government. Source of Truth AB is an equal-stakes joint-venture between Oxide AB and Robertson Nordic Partners AB.

Oxide has decades of experience building AI systems scaling to +100M users in production grade, recently solving adjacent problems for enterprises in banking, health and commerce.

Robertson Nordic Partners are behind the Nordic Game conference, long-view games ecosystem builders, and repeat government contractors for national, Nordic and European level industrial development.



Source of Truth AB, Box 20, SE 239 21 Skanör, Sweden, EU VAT ID SE559533810301

Please click here to contact us for more information.