Databiomes Unveils New Platform for AI Agents Running Locally on Consumer CPUs to Empower Game Developers and Enrich Gameplay
The Databiomes Platform Delivers a Cost-effective and Beneficial GenAI Solution that Amplifies Developer Creativity with No Recurring Costs and Minimal Latency
The Databiomes Platform is the result of intensive study of AI’s potential to make gaming more accessible and immersive, offering a proprietary inference engine that doesn’t require the cost and latency overhead and privacy concerns associated with cloud-based LLMs and SLMs. By providing on-CPU AI agents there is minimal overhead compared to on-GPU AI agents where 3D is heavily reliant. The platform provides an affordable, practical, and creativity-enhancing solution for game developers ranging from small studios to massive corporations. Pricing starts at $100 per AI agent for indie devs, with affordable enterprise pricing options for larger developers based on how many agentic functions the customer wants the AI agent to be trained to accomplish.
“We have pioneered AI running locally on the CPU because AI running on the GPU will impact graphical performance, which is a precious resource for both developers and players,” said Databiomes CTO Tomasz Klempka. “By utilizing the CPU and a small amount of RAM we can run multiple models with little to no impact on framerate, allowing graphically rich game experiences to benefit from robust AI dialogs, behaviors, and more.”
“Local AI models with a custom inference engine offer the best performance, something that’s critical for gaming and 3D workloads that typically have minimal headroom for additional features and processes,” said Steven Gans, CEO of Databiomes. “This use case inspired development of Databiomes Platform, which provides the AI benefits developers and players want without the cost and latency overhead of running models on the cloud.”
Databiomes preserves authorship and data privacy with a nano language model created exclusively from the customer’s input. From this, a model and tokenizer are trained from scratch within 24 hours, with the model and inference engine available for use under royalty-free license. From NPCs that talk naturally to players in real time to state machines that can boost development efficiency, the possibilities are immense.
.
Co-founder and CEO Steven Gans is a successful entrepreneur with a keen understanding of AI who has overseen multiple successful startups. A veteran of AMD, Arm, and Intel, he was formerly AI Product Manager at Intel, giving him depth in AI innovation at large scale, and CEO for Unreal Data, which was eventually owned and operated at IBM under his guidance. Tomasz Klempka, co-founder and CTO, is an Intel veteran known for optimizing hardware and software for AI to deliver the best results. He is passionate about using small language models to simulate and fully scale game worlds locally without additional costs and overhead. This vision inspired Tomasz and Steven to create the first on-CPU Super Resolution and Frame Interpolation AI model for Intel. This work led Tomasz to eventually prototype the first memory efficient on-CPU AI language model that became the Databiomes Platform.
.
According to Gartner Research, over 60% of AI initiatives don’t reach production due to bottlenecks such as performance variability, edge constraints, and integration complexity. Databiomes solves performance issues through partnerships with industry giants AMD, Arm, and Intel to ensure it runs smoothly and efficiently on all major CPUs and NPUs. This creates an integrated hardware and software design emphasizing speed, reproducibility, and scalability. Edge constraints are eliminated by Databiomes Platform’s inference engine, and integration complexity headaches such as tooling and framework issues are resolved through the single platform.
.
More information can be found at the official company site at databiomes.com. Follow Databiomes on LinkedIn and YouTube for the latest updates.
.
###
.
About Databiomes
Databiomes was co-founded in 2024 by AI on-CPU experts CEO Steven Gans, formerly of AMD, Intel, and IBM, and Tomasz Klempka, who engineered the first CPU-based Super Resolution and Frame Interpolation AI model for Intel. Headquartered in Toronto with global remote operations, Databiomes has leveraged $1M in pre-seed funding to pioneer an innovative platform for creating real-time on-CPU AI agents for game developers to enhance game experiences and production.