For developers
Build locally. Reach the network when you need more.
Rutli is being designed so applications can live close to the user and their private data, then call on participant-owned network capacity for workloads that belong there.
Local first. Network when useful.
A new deployment target.
The long-term ambition is simple to describe: build once, run locally or across a participant-owned network. Developers should eventually be able to reason about everything on the right without designing their application around a single hyperscaler from day one.
- local inference
- external inference
- batch workloads
- hardware requirements
- resource policies
- metering
- job status
- application distribution
- network capacity
Why applications matter to the network.
Applications do something a compute marketplace cannot: they create reasons for the node to exist.
- Applications create nodes.
- Nodes create supply.
- Supply makes applications more capable.
- Capability attracts more users.
- Users create more demand.
Software can become infrastructure distribution.
What you bring.
Applications on the node
More reasons to install, more nodes online because their owners use them. Your app is an engine of the network, not a customer of it.
Workload demand
External demand is what pays a node owner the dollar. First-party applications alone cannot fill a network.
Workload classes
The scheduler is only as useful as the kinds of work it can run interruptibly on heterogeneous machines.
Trust
Verified publishers, signed artefacts, reviewed workloads.
Keep sensitive work local.
A Rutli application should be able to distinguish between work that belongs on the user's own machine and work that is safe and appropriate for network execution. That boundary should be architectural, not merely a privacy checkbox in marketing copy. A financial record does not need to become a network payload simply because a remote GPU exists.
What is available now.
- Local application environmentWorking local AI applications and self-hosted systems.LIVE
- Network orchestrationFounding-node coordination and workload testing.PLANNED
- Public developer APIExternal developer access once the network contract and security model are stable enough to support it.PLANNED
- Distributed inference and trainingNot a current commercial product.RESEARCH
Help make participant-owned compute boring to integrate. That is when it becomes useful.