AI Research Lab · Reno, Nevada

Make the latent visible.

L8NTLABS is a small AI research lab run by Tom Duffy. I help businesses figure out where AI actually fits, build the custom solution, and make sure your team understands how it works. No hype, no jargon.

Start a conversation → See the work
[01]What I do 06 services

Custom AI work, explained in plain language.

01 Agentic Search Optimization Get found in AI search. As people ask ChatGPT, Perplexity, and Google's AI answers instead of scrolling a list of links, I make sure your business shows up in the response. → 02 AI Strategy for Small and Medium Businesses A straight answer on where AI helps your business and where it does not, with a plan you can actually follow. → 03 AI Coding Consulting Help your developers get real value out of AI coding tools like Claude Code and Cursor, without the hype. → 04 Agent Development and Deployment Custom AI agents that do real work, from first prototype to running in production. Built to be reliable and to fail safely. → 05 Privacy-Critical Implementations AI for sensitive data that cannot leave the building. Models that run locally, on your own hardware, so nothing is sent to a third-party API. → 06 Research Consulting A focused investigation into a hard AI question your team is stuck on, delivered as a clear, practical answer. →
[02] Research all projects →
in progress 2026.04.01

OpenArms

An open source library and SDK that gives AI models access to the physical world through microcontrollers and sensors.

View project →
in progress 2026.02.01

Orin

An agentic harness built specifically for smaller local models, able to write code and run scheduled tasks. In development, not yet public.

View project →
ongoing 2025.08.01

Mate1

A chess engine written from scratch in Rust, playing live on Lichess at around a 2076 Blitz rating and under active development.

View project →
[03] Blog all posts →
Local AI 2026.06.21

Switching From Ollama to llama.cpp For Local Inference

Local agentic coding is finally good enough to start taking architecture and performance more seriously. By switching from Ollama to llama.cpp, everything has improved for me. Speed, tool-calling, reliability...everything.

Read →
Local AI 2026.06.11

The "unlimited" AI plan was always going to end

The big AI companies are shifting from flat "unlimited" subscriptions toward paying for every token you use. Here is why that was inevitable, and why running your own models and tools is the best protection against being priced out.

Read →
[04]Contact no deck required

Have a problem that doesn't fit a template?

Tell me what you're building. I'll write back honestly about whether, and how, I can help.

Start a conversation →