built & building
- VOX OSdesigning
- CODING PLANNERbuilding
- VOX CANVASbuilding
- GOAL MAP / AGENT ORCHESTRATORbuilding
- TERMINAL ORCHESTRATOR (VOX TORCH)building
- DISCOVERY COPILOTbuilding
ADVANCED RESEARCH LABORATORY
AR LABS
an ecology of mind & machine
AR Labs is Richard M. Thompson’s AI engineering lab. We build AI agents that follow a business’s methods, with a person in charge.
-
-
01.1SALESBOTcurrently active — talk to it live
-
-
the ideas the lab runs on
-
02.1VOXseamless voice, generative visuals — a new way of working
-
principles & frameworks that inform the work
-
-
how the lab builds software with AI agents
-
packaged know-how the lab’s agents load
-
what the lab looked into, and what it found
-
the lab’s methods applied to a real problem
-
how one person and a set of AI agents run the lab
-
step by step: automating the routine work of a small business
-
essays from the lab bench
-
10ABOUTAR Labs & Richard M. Thompson
Start here
Running a business
- Talk to our voice agent
It runs a short sales discovery conversation, out loud, in your browser.
- Read how we would automate a music school’s bookings
A design case study, from first enquiry to a booked weekly lesson. Then the step-by-step guide.
- Meet Richard and book a call
Who runs the lab, what it offers, and a time to talk.
How it is built
- See how we build software with AI agents
For engineers reading the proof: specs, test-first builds, a second model’s review and a person’s merge.
- See what is actually built
All 33 projects with the status each one states, and where the source is.
- Open the goal map
The month’s goal, with the agent sessions working towards it.
Curious, or an AI agent
- Read what we believe about AI
The lab’s positions, written as views rather than forecasts.
- Browse everything
All 167 pages in one register, filterable by section and status.
- See the whole site as a map
Every section and page drawn as a circle by VOX Canvas, the lab’s own tool. Click a section to bring it to the centre.
- Read the site as an AI agent would
llms.txt lists every page; llms-full.txt holds them all as text.