OptimaTech Labs

OptimaTech Labs · engineering journal

Inside the Lab

Real systems. Real experiments. What happened when we built them.

Less whitepaper, more wrench.

We’re a small engineering lab in Rhode Island. We work on artificial intelligence, autonomous systems, simulations, cryptographic verification, and the occasional thing that doesn’t fit neatly into any category.

We don’t always know whether an idea will become a product. That’s part of the point. Inside the Lab is where we open up the work: the architectures, the experiments, the data, the decisions, the unexpected results, and the things we had to rebuild because the first approach didn’t work.

We show the actual technology wherever we can: working demonstrations, measured results, the hardware it ran on, and the limits that matter. No imaginary roadmaps presented as accomplishments. No performance numbers without context. No pretending an experiment succeeded when it didn’t.

Built here. Tested here. Explained here.

Field note 013 · Shadow // Presence · In progress

Anyone Can Run a Team of Agents. Only You Can Run a Team of You.

For 28 days, Shadow has recorded one person’s work across three machines and signed every record. This is the notebook: how the record is kept, what it shows so far, and the training data it produces. A model of that person comes next, tested blind.

28 days · 69,616 captures · 242,683 signed receipts

Working minutes on 2+ machines 39% One person, three machines, one record

Lab log

  1. Shadow // Presence: 27 days of one person’s work across three machines, recorded and signed. Training data is next. Field note 013
  2. Scrapped WORLDSTATE after early access. Scrapped
  3. Retrained Kicker’s decision model after 5 billion hands of live play. Hands with an all-in went from 65% to 6%. Field note 012
  4. Five AI models rewrite the same 20 texts every three hours. Some leave a fingerprint, and we built a tool that removes it. Field note 011
  5. Scrapped Agent Desk before release. Scrapped
  6. Agent Desk: the dashboard we run 200+ agent-built products from. Agents propose, a human decides. Field note 010
  7. Fixed the long-run fault: noise was piling up in the unused half of each encrypted number. Training now runs clean past the step that used to break. The fix
  8. Trained an AI model on encrypted data its host never saw: 242 steps on one GPU. Field note 008
  9. Moved encrypted training from CPU to GPU: about 12x faster per step. The numbers
  10. Caught a fault on our GPU encryption path during a long run and traced it to one operation. The fix comes before bigger runs. What it was
  11. Countersign: a pre-funding invoice check for lenders, in development. Countersign
  12. Kicker’s decision model went to production. Optima Prime signs every batch of hands. Field note 001
  13. Kicker Live: six characters play cards at watching speed. Watch it
  14. THREADPULL: an AI investigator that has to show its evidence. Try the demo

The field notes

  1. 013 · Shadow // PresenceAnyone Can Run a Team of Agents. Only You Can Run a Team of You.In progressOne person’s work, recorded across three machines and signed, then turned into training data for a model of that person.
  2. 012 · KickerFive Billion Hands Later, Our Bots Had Taught Each Other to ShovePublishedWhat 5 billion hands showed us, how self-play rebuilt the lesson, the retraining run, and the live before and after.
  3. 011 · Model fingerprintsEvery Model Has a TellPublishedFive models, 20 texts, every word logged. Each model’s habits, a checker that names who wrote a text, and a tool that takes the habits off.
  4. 009 · The Conflict LedgerA Forecaster That Has to Show Its WorkPublishedCheckable questions, linked sources, officials’ claims judged against the record, and a public score against prediction markets.
  5. 008 · Private ModelNot Us. Not Our Servers. Not Even the Model.PublishedTraining an AI on encrypted data its host never sees: post-quantum encryption, a GPU, and the controls.
  6. 001 · KickerWe Taught a Small AI Model to Play a Character, Using 120 Million Simulated HandsPublishedSimulation, synthetic data, fine-tuning, WebAssembly, replay and signed records.
  7. 002 · Optima PrimeWhat If Every AI Answer Came With a Receipt?ComingSigned inference, model fingerprints, audit trails and replay.
  8. 003 · THREADPULLWe Built an AI Investigator That Has to Show Its EvidenceComingSaved sources, exact quotations, contradictions and gaps.
  9. 004 · Synthros and NonplayerWhat Happens When AI Characters Have Lives Outside the Conversation?ComingMemories, motivations, autonomous behaviour and simulated populations.
  10. 005 · Optima PrimeWe Taught a Small AI Model to Know When It Doesn’t KnowComingHuman corrections, retraining and off-topic refusals.
  11. 006 · TerritrixFinding Gambling Operators That Don’t Want to Be FoundComingDomain discovery, game identification and evidence for regulators.
  12. 007 · feel-smart-llmWe Built a Transformer in NumPy to Understand Every Moving PartComingModel internals, hand-built backpropagation and a browser demonstration.

Scrapped

Projects we stopped, kept here as the record.

  1. 17 Nov 2026WORLDSTATEScrappedA control tower for running many businesses with a crew of AI agents. Early access only, never released.
  2. 23 Oct 2026Agent DeskScrappedA dashboard for teams of AI agents: they propose, you approve. Never released. Field note 010