NuNet Agents Hub lets you deploy open-source AI agents and keep your keys, your control, and your portability — without the infrastructure overhead. Beta is opening to a small group first.
A few questions about where you are today. We use it to make sure early access goes to the right people first.
No date set. Beta opens when it meets our own bar, not a deadline.
You wanted an agent running. Instead you got a server to babysit. Most builders stall in the same three places.
Sizing a box, SSH, ports, restarts, monitoring. None of it is the agent. All of it is in the way.
API keys pasted into config files and copied across half-finished setups. Nothing isolated, easy to leak, a chore to rotate.
Model spend is invisible as it runs. You find out when the credits are gone.
Start from an open-source harness. Hermes is deployable today, with more harnesses arriving through beta.
Choose the model that fits your task. Bring your own key or use managed access, and switch any time.
Keys handled, cost in view, state kept. A running agent, not a configuration project.
Beta is where the Hub becomes the place you run your whole agent operation, not just a single deploy. Here's what's landing. None of this is live yet, it's what we're building toward.
Pick an agent, a model, and where it runs. No server to provision, no SSH, no config to wrestle.
Several agents on one job, each with a role, handing off in sequence: one gathers, one drafts, one reviews.
An integrated storage layer: upload, browse, download, and export your agent's data. Your context stays live and your outputs are yours to access and move.
Models across providers. Bring your own API key or use managed access, and switch any time.
Credentials encrypted and managed for you. No secrets sitting in plaintext config.
Rules around the agent: scheduled context refreshes, backups, triggers. Its lifecycle on autopilot.
Give an agent its own HTTPS endpoint, callable from your app or automation tools. On request.
Update credentials, swap config, change skills on a running agent. From anywhere, no command line.
See what your agents are spending as it happens. Usage visible before the bill lands — no invisible token burn, no end-of-month surprise.
Beta opens when the deploy path is solid enough that we'd run it ourselves — not to hit a deadline. Early access is how we get there: a small group, real use, and we build from what we learn.
Pick your harness, pick your model, pick your machine, and swap any layer. The opposite of one stack you don't control.
Hermes today, more open-source harnesses through beta. Not married to one vendor's stack.
Managed for you, never left exposed across a DIY setup.
Your config, state, and data are yours to export and move. No lock-in, nothing lost on shutdown.
Cloud, your own machine, the edge, or a sovereign enclave. Same agent, your choice of ground.
Early access is going first to the builders and teams with real agents in play.
Early access opens beta to a small group first. Tell us what you'd run and where it hurts today, and we'll get you set up. Two minutes, and we read every one.
