The AI Workbench banner for Roger's Rabbit Hole: the RRH rabbit in a red apron installs a graphics card in an open PC tower whose power cord runs into a glowing rabbit hole
Welcome To The AI Workbench

The AI Workbench

Agents, local AI, open models, the platforms worth paying for and the projects worth stealing ideas from. Tested on my own bench, mistakes included.

I've been in tech since 1995, and I've watched plenty of "this changes everything" come and go. Lately, most of what's on my bench is AI. I run my businesses with a staff of AI agents. I host my own AI on hardware I put together myself. And I keep a pile of open models downloadable for anybody who wants them.

Most of what you read about AI is a demo, a sales pitch or both. This section is the other version. I tell you what I tried, what it's actually good for, what it costs and what broke. When something doesn't live up to the hype, I'll say so. When something does, I'll say that too, and if I can make a buck off a link, I'll tell you before you click.

What's on the bench

Agents at Work

My businesses run with a staff of AI agents, each with one job, a written persona and a list of things it's never allowed to do without me. Yes, they're named after characters from HBO's Silicon Valley. Between them they cover writing, art, video, research and operations. The one rule: agents propose, I approve. Nothing gets published, emailed, bought or deleted until I say yes myself. This is also where you'll find Field Notes, short real stories from the work, mistakes and all.

Local AI

CyberTron is my homebuilt local AI server. It started as a $150 computer and turned into a $592.33 workstation. It runs open-source tools like Ollama and Open WebUI, plus the Kokoro text-to-speech voices that narrate our YouTube Case Files. I picked that narrator blend by ear, and there's no per-minute voice bill. The whole build story lives in Building CyberTron.

Open Models

Hugging Face is the big hub for open models, and I pull a lot from it. I also help keep them downloadable through PirateFace, an outside project that turns eligible open models into torrents. My agents download and verify the files, and I run every submission myself.

Platforms

Which tool is actually for what, compared from my own use: agent platforms, local versus paid voices, model hosts and the rest. If I haven't used it, it doesn't get a verdict.

Projects

Code worth a look, mine and other people's. My own is the CyberTron Agentic Stack, a local-first agent memory project I forked from Agentic Stack and extended. It's public on GitHub. Worth a Look roundups tell you plainly whether I ran something or just read about it.

Tools and Products

The hot new stuff, put through a Bench Test: what it is, what I tested, what worked, what didn't, who it's for and what it costs. Some of these I may recommend or resell. When that's the case, you'll see it right at the top, and it doesn't change the verdict.

Start here

Nobody paid for this section. I use Grok Bot, Hugging Face and plenty of other tools, but I'm not a spokesman for any of them, and nothing here is official from xAI, X, SpaceX, Cursor, Hugging Face, GitHub or anybody else.

Latest From The Workbench
GPU VRAM: The Real Currency of Local AI

GPU VRAM: The Real Currency of Local AI

Graphics cards get sold on gaming speed. For local AI, VRAM is often what decides what you can run. Here's how to budget it: weights, KV cache, context, concurrency, quantization and multi-GPU, with CyberTron's RTX 3060 12 GB as the case study.

Monday, November 16, 2026/Author: Roggie D/Number of views (16)/Comments (0)/ Article rating: No rating
Building AI on Aging Hardware

Building AI on Aging Hardware

Useful AI doesn't always start with a brand-new workstation. Here's how a roughly $150 AMD FX-era desktop and a 12 GB RTX 3060 became CyberTron, a working local AI server, what it cost, how fast it ran and where the old hardware still slows it down.

Monday, November 16, 2026/Author: Roggie D/Number of views (33)/Comments (0)/ Article rating: No rating
Bigger Isn’t Always Smarter

Bigger Isn’t Always Smarter

There’s a belief floating around the AI world that the biggest model is automatically the smartest one in the room. More parameters, more gigabytes, more brains. It’s a nice, simple story. It’s also wrong.

Saturday, November 14, 2026/Author: Roggie D/Number of views (41)/Comments (0)/ Article rating: No rating
Building CyberTron — Running LLMs Locally

Building CyberTron — Running LLMs Locally

Testing Qwen models, GPU inference, and the software stack that finally gave CyberTron a brain.

CyberTron finally moves beyond hardware upgrades and starts doing what it was built for: running local AI. In this chapter, I put Ollama, Open WebUI, and the RTX 3060 to work, test Qwen models up to 14B, and see how far a $150 AMD FX-era machine can really go as a private AI server.

 

Tuesday, August 25, 2026/Author: Roggie D/Number of views (127)/Comments (0)/ Article rating: No rating
Building CyberTron — So What Did the $150 AI Server Really Cost?

Building CyberTron — So What Did the $150 AI Server Really Cost?

A full breakdown of the upgrades that turned a $150 computer into a $592.33 local AI workstation.

What started as a $150 used PC has grown into a $592.33 local AI workstation. In this installment of Building CyberTron, I break down the real out-the-door cost of every upgrade, including RAM, SSD storage, a new power supply, and the RTX 3060 12GB GPU. I also look at the hidden value of reusing displaced hardware across multiple workstations and answer the question every budget build eventually faces: did I actually save money, or just spread the cost across a series of questionable decisions?

 

Saturday, August 22, 2026/Author: Roggie D/Number of views (130)/Comments (0)/ Article rating: No rating
Building CyberTron - From Spinning Rust to SSD

Building CyberTron - From Spinning Rust to SSD

Moving CyberTron to solid-state storage without rebuilding the Linux environment from scratch.

CyberTron’s next upgrade replaces the old mechanical system drive with a Patriot P210 512 GB SSD. This chapter covers the storage migration, repurposing the original 1 TB hard drive for data, and how separating system storage, local data, and NAS backups made the $150 AI server feel more like real infrastructure.

 

Saturday, August 15, 2026/Author: Roggie D/Number of views (139)/Comments (0)/ Article rating: No rating
Building CyberTron — Why I Bought an RTX 3060 in 2026

Building CyberTron — Why I Bought an RTX 3060 in 2026

How 12 GB of VRAM turned an old AMD FX desktop into a much more capable local AI server.

CyberTron takes another major step forward with an upgrade from a GTX 1060 6 GB to an RTX 3060 12 GB. This article looks at why VRAM matters so much for local AI, why an older 12 GB GPU still makes sense in 2026, and how the upgrade opened the door to larger models, bigger context windows, and 100% GPU inference on a $150 AMD FX-era machine.

 

Thursday, August 13, 2026/Author: Roggie D/Number of views (138)/Comments (0)/ Article rating: No rating
Building CyberTron — The $25 RAM Upgrade From Hell

Building CyberTron — The $25 RAM Upgrade From Hell

How a simple 16 GB memory upgrade turned into a full hardware troubleshooting session.

What should have been a simple $25 RAM upgrade turned into a full-blown troubleshooting session. In Chapter 2 of Building CyberTron, I take the old AMD FX system from 16 GB to 32 GB of DDR3, fight through memory detection issues, test DIMMs and slots, update the BIOS, and learn once again that cheap old hardware rarely gives up without a fight.

 

 

Wednesday, August 12, 2026/Author: Roggie D/Number of views (130)/Comments (0)/ Article rating: No rating
Building CyberTron - It Started With a $150 Idea!

Building CyberTron - It Started With a $150 Idea!

How a cheap Craiglist desktop became the starting point for a private local AI server experiment.

Before CyberTron became a working AI server sitting in my lab, it was just an idea I kept coming back to: how much useful artificial intelligence infrastructure could I build without spending thousands of dollars? After a few weeks of talking through the concept, researching hardware, and debating what was realistically possible, I finally pulled the trigger on August 4, 2026. The cost of admission was just $150. That purchase officially started the CyberTron build journey and I didn't even know it, yet.

 

Wednesday, August 5, 2026/Author: Roggie D/Number of views (145)/Comments (0)/ Article rating: No rating
A Sensory Revolution Today and Beyond

A Sensory Revolution Today and Beyond

Haptic Technology

The United States government's "Black Budget" refers to the funding allocated for classified and secret projects within various government agencies, including the Department of Defense, the Central Intelligence Agency (CIA), and the National Security Agency (NSA). This budget is not publicly disclosed, making it a subject of intrigue and speculation.

 

Thursday, November 2, 2023/Author: Roggie D/Number of views (684)/Comments (0)/ Article rating: No rating
RSS
12

 

Follow Us on FacebookFollow Us on XFollow Us on InstagramSubscribe to our Youtube ChannelSubscribe to our Rumble Channel