Kanverse GPU Borrow
Borrow GPU compute from another device, with permission
Start new thread

Thank you, Product Hunt — this is only the beginning for Kanverse

by
•

Thank you to everyone who discovered, supported, upvoted and followed Kanverse GPU Borrow during our Product Hunt launch.

We now have our first 13 followers here. That may be a small beginning, but for an experimental technology project, it means a lot to us.

What is Kanverse?

Kanverse is an independent experimental technology lab where we build and test ideas around distributed computing, AI agents, device collaboration, security and new ways for computers and phones to share capabilities.

GPU Borrow demonstrated one of those ideas:

A device doesn't necessarily need to own a capability to temporarily use it.

We demonstrated a real NVIDIA RTX 3050 Laptop GPU being exposed through Blink Bridge, discovered and requested by an AI orchestrator, explicitly authorized by the user, used for a bounded workload, verified, and then released.

Since then, our experiments have gone further.

Android ↔ RTX
We demonstrated an Android device participating in a workflow where a data block travelled through Blink Bridge, was transformed using an NVIDIA RTX GPU, returned to Android RAM, and was later requested again and verified using SHA-256.

Blink Bridge
We're developing a temporary connection layer through which devices, applications and eventually agents can expose and borrow specific capabilities with explicit permissions.

Agent authorization
We're experimenting with AI agents that can discover available capabilities and request actions — while keeping execution bounded by authorization rather than simply giving the AI unrestricted control.

Proof & consent
We're also exploring ways to make agent actions verifiable, permission-based and auditable.

And there is much more being built.

Our larger question is:

What becomes possible when phones, PCs, GPUs, applications and AI agents can temporarily work together as parts of one system?

That is what Kanverse is exploring.

Explore KANVERSE:

We'll use this forum to share new experiments, technical results, failures, improvements and demonstrations as we continue building.

Thank you for being among the first people to follow the journey.

— Kanverse

25 views

Add a comment

Replies

Best

Want to see what we're building in action? 🛠️

Some of our Kanverse experiments are also being developed and demonstrated through hackathons on Devpost.

Check out our Devpost projects and hackathon submissions:


You'll find working prototypes, technical demonstrations and experiments behind the ideas we're sharing here.

More are coming

 How do you decide which device or GPU should handle a requested workload?

 Kanverse separates discovery from selection. First, the bridge discovers available GPUs/providers and normalizes their capabilities. Then the workload declares what it needs — for example CUDA support, minimum GPU memory, or a preferred GPU class.

The matcher filters out devices that do not satisfy hard requirements, then scores the remaining candidates. In the current prototype, CUDA compatibility gets the strongest weight, followed by preferred GPU type, available adapter memory, and a small preference for local hardware.

So for our public test, the request required CUDA and at least 3 GB of GPU memory. The integrated AMD GPU was not suitable for that requirement, while the NVIDIA RTX 3050 matched and was selected automatically.

The longer-term design is to extend the same decision with factors such as current load, latency, temperature, cost, privacy, and whether the capability is local, peer-to-peer, or cloud-based.

 That makes a lot of sense. I like how the hard requirements filter the options first, then the scoring handles the selection. The future factors like latency, cost, and privacy will make this even more interesting.

Live for 30 minutes: test my RTX from your phone

I’ve opened a guarded Kanverse GPU Router demo.

Your phone requests a bounded GPU workload over the Internet, Blink Bridge checks the requirements, selects my NVIDIA RTX 3050, executes the job, returns the result, then releases the GPU.

Safety is active: rate limiting, queue cap, 80°C temperature guard, and automatic shutdown.

Try it:

Tap ROUTE + EXECUTE GPU WORKLOAD and tell me if it works.

#GPU #EdgeComputing #DistributedComputing #Kanverse