Shelby said it has entered private beta, with its first customer deployments already in place across enterprise AI, distributed 3D rendering, and real-world data processing. The project is now testing a more practical question inside live customer environments: what a data layer needs to do when compute becomes increasingly dynamic.
Shelby narrows its focus
When Shelbyserves was first introduced, the project described its goal as building a decentralized, high-performance data layer for Web3, designed to make data easier to access, verifiable, and more useful. Since then, the team said its direction has become more defined.
Based on testnet operations and discussions with developers, Shelby said it found strong demand from artificial intelligence and other compute-intensive workloads. Once compute is distributed across multiple providers, regions, and different classes of AI hardware, the link between data and compute becomes a central issue. That observation pushed the project to sharpen its positioning.
Private beta is built around production workloads
Shelby said the private beta is organized around actual customer workloads and production data. Each integration starts with the business scenario itself: where the data sits, how compute is called, which tools the application depends on, and which performance requirements matter most to the customer.
The team said those requirements are changing quickly as AI infrastructure continues to evolve, compute options are updated, and new workload patterns appear. The private beta is intended to let customers shape the product while it is still being refined. Shelby said that process is meant to reveal which needs are shared across customers, which belong only to specific use cases, and where the product roadmap should go next.
Three initial customers, three different models
The first customer group shows how different those environments can be. Teepin is building enterprise AI infrastructure using proprietary data and open-source models. Its work on Shelby starts at the data layer and is expected to extend from storage into broader AI and compute services.
Pictor Network is building distributed 3D rendering infrastructure. Its workloads can use GPU resources in different locations, which means stable data availability directly affects how efficiently those compute resources are used. Shelby said Pictor has already integrated the platform into its development environment.
PathPulse is building spatial intelligence systems from video captured in real-world settings. Shelby described that as another data-heavy scenario, where large volumes of information are generated in one environment but may need to be processed and analyzed elsewhere.
Together, those three customers give Shelby three distinct deployment settings: private enterprise data, distributed GPU compute tasks, and large-scale real-world data.
Where Shelby says it fits best
Across those cases, Shelby said it sees a common pattern. The platform is most useful when data and compute are no longer fixed in the same place, infrastructure keeps changing, and the data layer has to adapt at the same pace.
In that setting, Shelby said it can reduce work tied to data replication, temporary caching, and synchronization. It also aims to simplify compute calls across different environments, allowing development teams to choose infrastructure based on workload needs instead of being constrained by where the data was originally stored.
The private beta is also meant to test the boundaries of that model and identify what additional features, integration interfaces, and tooling customers need around it. Shelby said that should help it plan the product roadmap around real shared demand rather than assumptions.
More deployment details are expected later
As those customer projects continue, Shelby said it will disclose more information over time, including how teams are using the platform, the scale of data involved, the changes seen after integration, and measurable implementation results.
The team said that process should make it easier to see which scenarios Shelby is best suited for, what supporting capabilities customers need, and where the product creates value across different applications.