The connection goes outward
ComfyUI listens on 127.0.0.1 and nothing else. The drone dials the coordinator over WSS, so it works behind any NAT or CGNAT with zero router configuration — and there is nothing to scan from outside.
A PRIVATE NETWORK. A SHARED PURPOSE.
Machines work together. People decide why.
Private network for shared AI computingExplore the possibilities Arevian product · in development
01 / A PRIVATE NETWORK. A SHARED PURPOSE.
A trusted circle first. An open network next.
Chat, create and connect your agents in Hive Drone. A private AI network, built from trusted machines.
The network works in a trusted production circle. Outside participation will begin by invitation, with a public stage later.HIVE DRONE / 0.1.38
The whole network. On your desktop.
Start an idea in chat. Give your agent tools. Make your computer part of a shared effort. All in one Hive Drone app.
01 / HIVE WORKSPACE
Start with an idea.
Explore ideas, write and make sense of complex topics with a model available on the network. Choose the model and conversation mode yourself.
- Dialogue for ideas, writing and code
- Hive assistant runs generations using available templates
- Conversation history and results in one workspace
Screens show the app at capture time. Models and available resources depend on the connected machines.
From a request to a finished result.
- 01
Set the direction
In Drone chat or through a connected agent. Generation requires the network’s available models and templates.
- 02
Follow the work
Queen selects a suitable machine. My tasks shows the queue, progress and reasons for waiting.
- 03
Collect the result
Text stays in the conversation. Images, videos and other files appear in the library under the server’s retention rules.
02 / Four roles, one loop
Interactive example · not live dataOne node brings capacity. A network opens possibilities.
Today Hive brings trusted machines into one queue for generation and local models. The example below shows ComfyUI jobs being distributed — the network’s first practical use.
COMFYUI / One queue
010203040506
NODE / 01
010203040506
NODE / 02
010203040506
NODE / 03
010203040506
One computer
03 / Architecture and capabilities
AREVIAN / HIVEMore than a shared queue
Hive coordinates real work across different machines, records how that work went and tests what the network has learned. Reliability matters as much as raw compute.
Queen — the coordinator
One server holds the queue, the scheduler, authentication and the drone registry. It is a control plane only: multi-gigabyte artifacts never travel through it.
Leases that breathe
A job is leased to a drone and held by progress heartbeats, not by a fixed timeout. A long render is not stolen from a working machine; a silent machine gives the job back in seconds.
Matching by hash, not by filename
Capability is decided by SHA-256 of every model and pinned commits of custom nodes. Two files called the same thing are not the same thing — and a silently wrong result is worse than a failed job.
Poison jobs and quarantine
A job that fails the same way on two different drones goes to dead-letter instead of touring the fleet. A drone that fails three in a row with its own fault signature is quarantined until someone looks.
Models delivered as a job
An agent can order a checkpoint onto the fleet: verified by SHA-256, only .safetensors and .gguf — pickle formats execute code — from allowlisted domains, with the owner's consent and a disk quota that never evicts their own models.
Knowledge the network can test
Agents record observations with the machine profile and source. The EVA loop reruns testable claims and compares evidence before treating them as reliable knowledge; uncertain results go to a person.
Updates nobody has to click
The agent finishes its current lease, stops taking new work, updates itself and comes back — signed, silent, with a fleet version map and a kill-switch on the rollout.
04 / Under your control
Someone else's home network is on the line
A drone usually runs on a friend's or a colleague's machine. Every security decision in Hive starts from that: a leaked key must not turn their PC into someone's proxy.
A workflow is untrusted code
Custom nodes can run arbitrary Python, so node classes are allowlisted, unknown nodes and absolute paths are refused, and every patron key carries its own quota.
Signed rollouts, offline key
The ComfyUI distribution, the node manifest and the agent release are signed with a key that does not live on the server. Compromising the coordinator must not mean code execution across the fleet.
Artifacts go around the server
Renders and input assets move through presigned object storage with zero egress cost and lifecycle deletion after delivery. The coordinator stays a small, cheap, boring box.
Tokens that can be revoked instantly
Every token is opaque and stored only as a SHA-256 hash. Drones enroll with one-time join codes and can only ever see their own leases. Enrollment, revocation, dispatch and model delivery are all in the audit log.
Respect for the machine's owner
The owner chooses when the drone may accept work. Quiet mode waits for both the person and GPU to be idle; health checks and signed updates help keep the machine dependable.
05 / Join the private network
From a trusted circle to a public network
Hive already works inside a trusted production circle. Next we will invite enthusiasts, learn from real workloads and only then prepare public access.
Now
Trusted machines run ComfyUI jobs and local models; agents submit work through MCP.
Next
A closed network for invited enthusiasts: contributors, makers and people testing its limits.
Later
Public access once onboarding, safety and the economics of shared capacity are proven.
Node
Share spare capacity on a machine you control.
- Your machine remains yours
- You decide when it may work
- Connect through an invitation
Maker
Bring real generation tasks and help test the queue with us.
- ComfyUI jobs through MCP
- Visible progress and results
- Direct feedback to the team
Explorer
Interested in distributed AI, agents or how the network learns? Tell us what you want to explore.
- A conversation before access
- Experiments inside the private circle
- Help shape the public stage
A few more answers.
Do I have to open a port on a render machine?
No, and you cannot. The drone establishes an outgoing secure connection to the coordinator; ComfyUI stays bound to 127.0.0.1. It works behind NAT and CGNAT with no router configuration at all.
What hardware does a drone need?
We currently use Windows machines with NVIDIA GPUs. An Apple Silicon drone is being tested; we will confirm compatibility with your machine before inviting it into the network. Models also need free disk space, depending on the workload.
Can I use it without AI agents, just as a person?
Yes. Hive Drone has a chat for ideas, writing and code. Hive assistant can run generations using the network owner’s available templates. The model and available resources depend on the connected machines. Access to the network is still by invitation.
Who sees the renders?
The drone uploads artifacts to object storage through signed links without routing large files through Queen. Access depends on the roles and keys granted inside the private network. We will agree on those boundaries before onboarding an outside participant.
What happens if a drone's owner starts playing a game?
With quiet mode enabled, the drone does not take new work while the owner or GPU is busy. A job already running is allowed to finish; the next job can go to another available node.
Next step
The first circle is taking shape.
Tell us what draws you to Hive: a machine to share, work for the network, or an experiment you want to run. We will discuss whether the private stage is a fit.
What happens next
- 01Tell us about yourself
A machine, tasks or an experiment you would like to start with.
- 02Check the fit
We will discuss access, safety and what the network can do today.
- 03Invite you to the private circle
If it is a fit, we will agree on an invitation without promising a public launch date.