Shelby's Next Stage Starts With Real Workloads

When @shelbyserves was first announced, the vision centered on a decentralized, high-performance data layer for Web3 - infrastructure that could make data more accessible, verifiable, and valuable.
Since then, the opportunity has become more specific.
Testnet and conversations with builders surfaced particularly strong demand from AI and other compute-intensive workloads. As compute spreads across providers, regions, and different types of AI hardware, the relationship between data and compute becomes increasingly important.
That has sharpened Shelby's focus.
Private Beta is now live, with the first customer workloads running across enterprise AI, distributed 3D rendering and physical-world data.
This stage brings Shelby into real customer environments and starts answering a more practical question: what does the data layer need to do when compute is increasingly dynamic?
Start With the Workload
Private Beta is built around customer workloads and production data.
Each engagement starts with the workload: where data lives, how compute is being used, which tools the application depends on, and what requirements matter most for that customer.
Those requirements are also shifting quickly as AI infrastructure evolves, compute choices change, and new workload patterns emerge.
Private Beta gives customers a chance to help shape Shelby around those needs as they develop. It is deliberately a co-build: a way to see which requirements repeat across customers, which are specific to individual use cases, and where the roadmap should move next.
Three Workloads, Three Different Data and Compute Patterns
The first cohort reflects that range.
Teepin (@teepin) is building enterprise AI infrastructure around proprietary data and open-source models. Its rollout onto Shelby starts with data, creating a path from storage into broader AI and compute services over time.
Pictor Network (@pictor_network) is building infrastructure for distributed 3D rendering. Its workloads can draw on GPU capacity across different locations, making data availability an important part of how effectively that compute can be used. Pictor has already integrated Shelby into its development environment.
PathPulse (@PathPulse_AI) is building spatial intelligence from video captured in the physical world. That creates another data-intensive pattern, where large volumes of information are generated in one context and may need to be processed and analysed elsewhere.
Together, the three customers give Shelby very different environments to work against: proprietary enterprise data, distributed GPU workloads and physical-world data at scale.
Where Shelby Fits Best
Across those use cases, a common pattern starts to emerge.
Shelby is most useful when data and compute do not stay in one place, and the data layer needs to keep up as infrastructure changes.
That can mean reducing the work around copies, staging and synchronisation. It can mean making it easier to use compute in another environment. And it can give teams more freedom to choose infrastructure based on the workload rather than where the data already sits.
Private Beta will help show not only where that pattern holds, but which additional capabilities, integrations and tooling customers need around it. That gives Shelby a way to shape the roadmap around recurring customer needs rather than assumptions.
What Comes Next
As these customer engagements progress, Shelby will share more about what each team is using the platform for, how much data is involved, what changed once Shelby was introduced, and the results that can be measured.
Over time, that will give a clearer picture of where Shelby works best, which capabilities customers need around it, and why it matters across different use cases.
Your compute strategy is still evolving. Your data architecture shouldn't lock it down.
If that challenge sounds familiar, get in touch with the Shelby team or follow along via Discord.