Individuals / teams / applications

Build connections around your work.

Use an existing network, operate one for a group, or connect software through an explicit service interface.

These are practical ways to use the released capabilities. Each depends on the relevant network, host and access setup.

For individuals and small groups

A project group can publish reviewed reference Pages, exchange encrypted letters and share small files with recipient approval. Suite provides a native browser and dedicated Mail and Transfers windows, while terminal clients support environments without a graphical desktop.

Start by previewing a local example Page. Joining a live network then requires accepted trust, membership and routes from that network's operator. Reading Pages does not require hosting a network yourself.

Install Suite and try a local Page · Join an approved network

For organisations and teams

A team can maintain reviewed internal publications, collect form submissions and expose selected services. Membership, publication review, confidential content and application access remain separate permissions, so admission to the network does not grant every capability.

Operators choose hosting resources and retain audit and recovery records. Separate networks can agree to exchange selected Pages while keeping independent administration.

Plan the authority replicas, trust custody, clock health, service admission and recovery procedure before inviting users. HyperNet's deployment model needs explicit operation; it is not an automatically configured organisation network.

Prepare a network · Understand authority · Share reviewed publications

For application and agent developers

Give an application its own identity and grant access to a specific service. Use the local companion and Python SDK to open a relay-only stream, then exchange the application's protocol within the granted limits.

A local automation agent can use the same interface to request an approved operation without receiving a user's private keys. The application still owns its input validation, operational permissions, cancellation and recovery. HyperNet supplies the connection and peer context; the developer supplies the application behaviour.

Connect an application · Read the Python SDK · Explore examples

For connected FICC clusters

Two FICC clusters within an approved HyperNet network can pair, publish selected inference services and grant remote operations. Requests travel through HyperNet application streams, while the receiving FICC cluster enforces its own project grants and resource limits.

A team can expose an inference service at another site, or reuse a verified model through FICC's resumable file protocol. The model transfer uses verified ranges across finite streams. It does not depend on, or enlarge, HyperNet's native Transfers object limit.

Inference engines, model storage and GPU scheduling remain FICC responsibilities. HyperNet does not provide cross-host tensor parallelism or merge the clusters' administration.

For infrastructure operators

Offer selected hosting, relay or application services with bounded resources and explicit consent. Retain control of the machine and withdraw its contribution when needed.

Choose the operator relationships to match the intended privacy boundary. A route through infrastructure under a single operator's control does not establish the same privacy assumptions as independently operated infrastructure.

Hosting and resource control · Route privacy · Operational recovery