Live now: 0 tokens processed in the homelab

The operations lead who builds his own tools.

Cameron Barry turns messy operations into systems that run, lately with AI doing the heavy lifting. Thirteen years in production across telecom, SaaS, ecommerce and wealth management.

A demo is an interview. The spec sheet is a resume.

Field-tested at

Lethal Gaming GearPulsarCCI SystemsRev.ioAdvizorProAT&T

What you get

Four capabilities, each with the receipts.

Operations from zero

Built Lethal Gaming Gear's operations before there was a warehouse: inventory, fulfillment, purchasing, vendors, hiring, payroll and playbooks. Then grew it through acquisition.

$2.8M

peak revenue, with about 100,000 orders in the peak year

Systems, built with AI

Designed and built Aura, an all-in-one warehouse, ERP and CRM system, after learning the work from the people doing it. Built with AI-assisted development instead of waiting on engineering.

$42K

a year in labor saved, estimated, across the U.S. warehouse

Programs that scale

Ran Comcast's regional 5G WiFi pilot with a team of 13 design engineers, then came back to turn it into a national deployment with the intake, tracking and QC to match.

500+

people supported across multiple markets

Integrations that fit the work

Configured CRM integrations across Salesforce, HubSpot, Microsoft Dynamics and GoHighLevel, and was the go-to REST/SOAP API resource at Rev.io. An integration only fits if you understand the business behind it, so he maps the system to the workflow, not the other way around.

13 years

learning how businesses work, across telecom, SaaS, ecommerce and wealth management

Release notes

Every version since the first one shipped in 2013, most recent first.

    How it works

    The same five steps, whether the problem is a warehouse, a rollout or a CRM.

    1. Ask a lot of questions

      Learn how the business actually works before touching anything.

    2. Find out why it's built that way

      Every process exists because someone was solving a problem.

    3. Decide: adjust or rebuild

      Sometimes it's a small fix. Sometimes it's a full rebuild. Knowing which is the job.

    4. Build it, often with AI

      Design the system, direct AI to build it, and test it hard.

    5. Document it

      So the next person, or the next rollout, starts from a better place.

    Live demo

    A tool built for a real salon, running here with made-up clients and replies. When someone cancels, it ranks who's most likely to want the slot, texts them one at a time, and books the first yes.

    The cancellation refiller

    Try it: cancel any appointment and walk the slot through to rebooked.

    Saturday

      1. Slot opens
      2. Rank clients
      3. Text one at a time
      4. Book the first yes

      System status

      Served live from Cameron's homelab, where he tests models, runs agents and breaks things on purpose.

      All systems operationalUpdated just now

      Tokens processed

      0

      Qwen, local: 0 Claude: 0

      Model loaded
      Qwen3.6 35B-A3B
      Running on
      RTX 2080 + Arc A380, i7-12700K, 48GB DDR4
      GPU load
      0%
      Now working on
      Nightly job pipeline review
      Uptime
      0 days

      Token counts are usage totals, not bills. Claude usage counts from when it joined the lab.

      Integrates with

      Works with the stack you already have.

      AI and automation

      • Claude Code
      • Anthropic and OpenAI APIs
      • RAG
      • llama.cpp
      • n8n
      • Zapier

      CRM and commerce

      • Salesforce
      • HubSpot
      • Microsoft Dynamics
      • GoHighLevel
      • Shopify

      Workspace

      • Notion
      • Microsoft Office
      • Google Workspace

      Systems

      • Windows
      • macOS
      • Linux

      Choose your edition

      Same product, configured for the role you're hiring for.

      TPM

      For teams that need business problems turned into technical work.

      • Translates between customers, developers and data teams
      • Scope, risk and sprint reviews from discovery to go-live
      • Hands-on with APIs and integrations

      Pricing: let's talk

      Request this edition

      Project Manager

      For rollouts that have to work at scale and survive delays.

      • Took a regional pilot to a national program
      • Intake, tracking, QC and delivery processes
      • Executive working sessions that unblock things

      Pricing: let's talk

      Request this edition

      Operations

      For growing companies whose processes are falling behind.

      • Built ops from zero and led an 8-person team
      • Vendors, inventory, fulfillment and hiring
      • Ticketing, automation and internal tools

      Pricing: let's talk

      Request this edition

      Frequently asked questions

      Is Cameron a software engineer?

      No, and he doesn't pretend to be one. He's been into AI for years, and working alongside engineers taught him what good work looks like. So he designs the system, directs AI to build it, tests what comes back against that standard, and ships it.

      What happens if there's no onboarding?

      Self-onboarding is included. At AdvizorPro there was no formal training, so he used Claude to mine Slack and email history, filled the gaps by hand, verified it against the live systems, and built a Notion knowledge base covering the whole integration role. He designed it as onboarding for whoever came next.

      Remote, hybrid or on-site?

      Remote anywhere in the U.S., hybrid in the Atlanta area, or relocation within Tennessee, the Carolinas or Florida.

      Why does Cameron run a homelab?

      Because the best way to understand AI in operations is to run it. The homelab benchmarks open models, runs agents, and powers the status page on this site. It's the same work as AI Ops at a company: picking models for the job, fitting them to the hardware, wiring agents together, and keeping it all running.

      Can I see the code?

      Aura was built for Pulsar, so its code is private. The AI job search pipeline, a self-hosted tracker with Gmail status tracking and nightly local-model reviews, is going public on GitHub soon.

      Scheduled maintenance

      Even the best systems need downtime. Here's where his goes.

      Evenings and weekends

      The barn and the ballfield

      Horseback riding for his daughter, baseball and basketball for his son.

      Ongoing

      His wife's salon

      He helps run it, and it's his favorite test bed. The demo above started there, along with a tool that watches supplier sites for sales on the products she uses most.

      Whenever the weather allows

      A 1995 ProCraft bass boat

      Rebuilt from the fuel system up, with a new dash panel designed in Fusion 360 from a 3D scan and laser-cut in aluminum.

      Request a demo

      Looking for the next TPM, project management, AI Ops or operations role. Tell me what you're hiring for and I'll get back to you.