This solo builder runs 24/7 local AI on his own hardware | Alex Finn

July 13, 2026

AI Summary

5 min read

Alex Finn runs a small fleet of AI computers in his office—three Mac Studios, a DGX Spark, and a custom-built RTX 5090 machine—and they are all burning tokens around the clock. The result is what he calls "ambient AI": models that never stop working, scanning code, monitoring social media, and building software without the per-token cost of cloud APIs. The point, he insists, is not pure ROI. "You now have, because you have AI models running locally, the ability to run unlimited intelligence around the clock 24/7. If you were to do that with a cloud model like ChatGPT or Claude, you would be spending outrageous amounts of money."

The hardware stack: matching machines to jobs

Finn runs four categories of hardware, each with distinct tradeoffs. Mac Studios offer high unified memory (up to 512 GB), which lets them run very large frontier models like GLM 5.2 locally. The catch is low memory bandwidth, so responses can take minutes. These machines are best for slow, high-intelligence tasks like deep security analysis. AI computers like the DGX Spark sit in a sweet spot: they have unified memory from NVIDIA (128 GB), decent bandwidth, and CUDA acceleration, making them good for mid-size models like Qwen 3-6 that need reasonable speed and intelligence. Traditional NVIDIA GPUs like the RTX 5090 have only 32 GB of VRAM but are lightning fast, delivering cloud-like speeds locally. Th

Continue reading the full summary in the app — free to try.

Read Full Summary →

Free • No credit card required

Listen to Audio Summary Open in App

Never miss an episode of How I AI

Get every new episode summarized in your inbox — free, ~5 minutes to read.

No spam. Unsubscribe anytime.

What you'll learn

1 (00:00) **Hardware Stack & Ambient AI Philosophy** - Alex Finn introduces his office setup: three Mac Studios (512GB each), a DGX Spark, and a custom RTX 5090 build, all burning tokens 24/7.
2 (03:13) **The "Red Pill" Moment: Discovering OpenClaw** - Alex traces his journey from a January blog post about OpenClaw to buying a Mac Mini and realizing he wanted a personal AI assistant living on his own hardware.
3 (07:22) **Hardware Breakdown: Mac Studio vs. DGX Spark vs. RTX 5090** - A practical comparison of the three main local AI hardware options, their strengths, and weaknesses.
4 (13:35) **Zero-Tech Setup: How to Install Models** - OpenClaw and Hermes agents can act as your IT guy, automatically scanning hardware and loading the appropriate model.
5 (17:31) **The "Unlimited Intelligence" Use Case** - The core value of local AI is running unlimited intelligence 24/7, which is cost-prohibitive with cloud APIs.
6 (20:44) **Federating Work: Local Models as SDRs, Frontier Models as Closers** - A practical division of labor between local and cloud models.
7 (24:29) **OpenClaw vs. Hermes: The Agent Showdown** - A candid comparison of the two leading agent frameworks.

+ Full timestamped outline available in the app

Guests on this episode

Alex Finn

Show Notes

Alex Finn is an AI builder, YouTuber, and the creator of Vibe Code Academy, a community for people learning to build with AI tools. He runs one of the most ambitious local AI setups I’ve come across: three Mac Studio 512 GB machines, a DGX Spark, and a custom RTX 5090 build, all coordinated through a fleet dashboard he built himself. He’s spent five months figuring out which local models belong on which machines, how to wire them to Claude Code loops, and how to get a software factory running without babysitting it.

What you’ll learn:

How Alex chose between a Mac Studio (512 GB unified memory), DGX Spark, and RTX 5090, and what each is actually good for
Why Tailscale is worth installing even on a single machine, and how it lets one agent manage your entire hardware fleet
How the build loop and review loop in Claude Code work
How to allocate tasks by machine and model
Why unlimited local inference changes the use-case math in a way a $20 cloud subscription never can
What OpenClaw and Hermes are each best suited for, and why Alex runs five agents total with failover baked in

—

Brought to you by:

Runway—The creative AI platform for images, video, and more

Jira Product Discovery—Prioritize with insights, build with confidence

—

In this episode, we cover:

(00:00) Intro

(02:58) Alex's hardware stack

(03:48) What "ambient AI" means

(04:15) Alex's red-pill moment with OpenClaw

(07:04) Mac Studio vs. DGX Spark vs. RTX 5090

(13:24) How to set up local models with no technical knowledge (Tailscale + OpenClaw/Hermes)