Marketing and Community Lead @nunet_global | Building @zeroarc_ai
Joined December 2024
- Tweets37
- Following12
- Followers41
- Likes19
I've had Kimi and GLM bookmarked to test for the longest time
Two open-weight models I kept meaning to actually run. Never did. Every time it meant standing up a server, pasting keys, and watching a bill I couldn't predict. It never made the top of the list.
This week it did.
I've been testing @Undergryd. It takes the infrastructure work out of running an agent.
I deployed a Hermes agent. Switched the model from inside the app; no developer, Telegram works too. Now I've got Kimi and GLM running side by side against my current stack.
No server. No Docker. No terminal.
Credits load up front and carry a ceiling, so there's no surprise invoice at month end.
The question I've sat on for a year: does an open-weight model hold up on the work I'm already paying frontier prices for?
Now I can just watch it happen.
If you run agents for clients, you know the tax. A box per client. Keys everywhere. A bill nobody saw coming.
Undergryd is opening early access to a small group of founding operators. If you've felt that tax, this is the moment to get in while the group is still small.
Want a look? Head to earlyaccess.undergryd.ai
Got a great catch from some users asking how the audit handles trackers that load AFTER the page first appears.
v6 shipped this morning. Four-pass dynamic fingerprint — initial / post-consent / post-scroll / synthesise. Catches the pixels GTM injects, the ones gated behind cookie banners, and the SPA-rendered content the static check missed.
Round 6 in the iteration log: github.com/jennifer509/herme…
This is why iterating in public > polishing in private
I've been experimenting with Hermes Agent on NuNet's compute network. Built a website-audit tool entirely through chat.
No code editor, no API keys, no setup scripts.
Three tricks I figured out, including one moment that genuinely surprised me.
Open-sourced the v4 skill + the full experiment log: github.com/jennifer509/herme…
Most AI agent content right now is screenshots of cool things. The harder content is showing how you build with them, including what they get wrong.
@nunet_global running an alpha for an agent platform built around this kind of operator setup. Sign up at community.nunet.io/agents.
If you've been experimenting with an AI agent, what's the most useful thing you've made it do?
Testing out 60 second AI Agent deployment on Decentralized Compute @nunet_global this evening 👀 #HERMES @NousResearch
If you can Docker it, you can @nunet_global it
Built a daily AI news digest — scans 250+ articles,synthesizes via Gemini, delivers to Telegram. Packaged it in a 6-line Dockerfile and deployed on decentralised compute
Full code + guide: github.com/jennifer509/nunet…
I built this deployment and recorded the walkthrough. What surprised me most is how straightforward the whole process was.
OpenClaw, Ollama, and Qwen 3.5 running together as a private AI agent, deployed through the NuNet Appliance in about five minutes. No cloud account, no API keys, no data leaving the machine.
The part that gets me is the private networking. @nunet_global creates a dedicated network for each deployment so the containers talk to each other internally. Your inference traffic never touches the public internet.
If you’ve been curious about running your own AI setup locally, this video walks through every step.
We deployed a fully private AI agent on NuNet in under 5 minutes 🚀
@openclaw running @Alibaba_Qwen through @ollama , one of the hottest open source model families right now, entirely on decentralized compute. No cloud. No API keys. No data leaving the machine.
This is what private AI looks like when you actually build it instead of just talking about it.
Your model. Your hardware. Your rules.
Full walkthrough showing exactly how it works: youtu.be/KXpoorhADuo
What should we deploy next?
VS Code and VirtualBox have both earned permanent spots in my dock.
6 months ago I didn't know what a terminal was. Now I run Claude Code from inside VS Code and it's where I do most of my work.
A few shortcuts that save me the most time:
Cmd + J — toggle the terminal open/close without touching your mouse
Ctrl + W — delete the last word you typed
Ctrl + X — clear the whole line and start fresh
Up arrow — re-run your last command instantly
Ctrl + C — stop something that's running (you'll use this a lot)
Cmd + P — find any file in your project in seconds
VirtualBox is there because I'm working on the NuNet network. Running nodes, testing infrastructure. That's a story for another post.
None of this felt natural at first. I Googled every shortcut. I broke things constantly.
But the gap between "I have no idea what I'm doing" and "I can actually build things" is smaller than you think
Jennifer Bourke retweeted
Thank you for the very interesting space, @KGeN_IO
It seems we have a conclusion of what the bread and butter for AI agents are:
compute + data!
You need both - and you need them healthy and clean.
For the compute side of things, ask your agent to explore NuNet 🤝⚡️⚡️
It will still need your help in setting things up, but it will be very happy to be able to operate in a really decentralized, peer-to-peer environment. If you have to pay anyone for compute, why not pay your web3 peers, rather than big tech? The space grows stronger this way.
@sanjaybuilds_ @RishAhuja @sparsh_noahAI @maeveknows @emetlalune @pukecast @RugRadio
🟩Your Next Coworker Isn’t Human.
Join us in an all-human ThoroughScan tomorrow as we discuss “AI Agents as the New Digital Workforce.”
Guest Speakers:
@sanjaybuilds_ @RishAhuja @sparsh_noahAI @maeveknows @emetlalune
@pukecast @RugRadio
📅 3rd March
⏰ 2:00 PM UTC / 10:00 PM HKT
Set your reminders 👇
What I like about this rollout: providers come first, the network grows with real participation, and it builds up from there.
Phase 1 is live.
If you've been curious about decentralized compute — this is a real way in.
NuNet Network Live is seeding — and here’s exactly how it works
The network is a multi-sided platform. We’re building it in stages, starting with the foundation: compute providers.
Phase 1 (Now) — Compute Providers: If you have a PC, server, cloud instance, Mac Mini, or even a Raspberry Pi, you can contribute compute to the network. Download the Appliance, onboard your resources (CPU, RAM, disk, GPU), and join NuNet Live. The team manually reviews every application and is providing hands-on support. You are an early user — your participation directly stress-tests availability, uptime, and performance across the live protocol. Compute providers are compensated in NTX for real work done on the network.
Phase 2 — Easing Access: As the provider base grows and early issues are resolved, we move toward automatic approvals and reduce manual checks. The goal is to make joining the network smoother and faster.
Phase 3 — Orchestrators: Real workloads come to the network. Selected orchestrators will deploy their applications on devices across the network.
Phase 4 — Organizations: The full vision. Organizations configure their own subnetworks with their own requirements, contracts, and governance — tailored to specific use cases like energy systems, robotics, data sovereignty, and multi-cloud orchestration.
In parallel: NuNet is rolling out applications and use cases developed over the past few years — both internal and with business partners. Announcements coming in the weeks ahead.
As @vveitas put it at launch: “The community takes the first seat in the evolution of the network. Your participation, your feedback, and your involvement will shape what’s coming next.”
We recommend using a dedicated device where possible, and as general advice, avoid using devices that contain sensitive personal or business data.
Setup guide → docs.nunet.io
Jennifer Bourke retweeted
CARDANO PAYMENTS NOW LIVE 💪
We've just demonstrated seamless Cardano blockchain integration for decentralized compute payments:
👉🏻 Real-world utility: Deploy and pay for compute jobs with $ADA
👉🏻 $NTX ecosystem: Utility token driving network orchestration
👉🏻 Seamless UX: Toggle between Ethereum and Cardano with one click
👉🏻 Production ready: Full payment flow from deployment to settlement
This represents another critical milestone as we approach our March 2nd live network launch.
The infrastructure is proving itself ready for global adoption. Multi-blockchain support, seamless payments, and real utility - all operational 🚀
youtu.be/mm2H2Edxid4