Context

Based on this vision, this doc outlines a Product and Engineering Roadmap of how we get there one milestone at a time.

In the vision, there are three core deliverables identified to ensure the success:

1. Ability for users to deploy realtime AI workflows and request inference.
2. Industry-leading latency for realtime AI and world model workflows.
3. Cost-effective scalability, allowing users to pay as they go and the network to automatically deliver required scale.

Different from traditional infra providers for the same space (media inferences), the Livepeer is uniquely identified with following values:

1. Open access and permission-less nature: As a protocol on a public blockchain, anyone can build on the network, reducing platform risk.
2. Open source and community oriented: The project is open source, allowing users to control, update, and contribute, and benefit from network improvements as token owners.

What is it: LP Network-as-a-Product?

What is it?

LP Network-as-a-Product is a service-oriented product that consists of three key components:

#1. At its core, it is the Permission-less Livepeer Protocol enabling Orchestrators to provide their GPU capacity as a service on the network.

as an orchestrator, I am able to enroll my GPUs to the Livepeer Network by knowing what SLAs it must comply with, what demands this GPU is to satisfy, and what compensation it pays.

#2. A set of public monitor-able real-time video inference SLAs that define what the network offers.

as users of the network, I am ensured by all my inference services requests will be met with a pre-agreed, and published SLAs.