PRODUCTS / OPERATE THE MACHINERY

Each product gets its own door.

Enter the business workspace, follow the autonomous market architecture, run the live experiments, or open the complete systems field. No product is buried in a homepage scroll.

Obtuse Work — one workspace, every capability below.

A local-first Windows workspace where routine work moves inside explicit policy and consequential authority stays with named people. Everything listed here ships in the current build — nothing is a roadmap promise.

Command & customers

  • Deterministic Command Queue ranked from every open record
  • One customer file across sales, money, schedule, and delivery
  • Exact-record execution: the queue opens the invoice, message, or approval that created the pressure
Open

Money

  • Quotes, invoices, and a reusable line-item library with exact-cent math
  • Bank statement import, duplicate protection, and human reconciliation
  • Recurring invoice drafts, PDF documents, receipts, and receivables aging
Open

Scheduling

  • Local drafts with conflict checks before anything is published
  • Operator-published fit-call windows and human-confirmed requests
  • Google Calendar create, cancel, and replacement behind explicit approval
Open

Communications

  • Signed end-to-end encrypted inbound voice, SMS, and web events
  • Consent, opt-out, routing, quarantine, and per-channel pause
  • Draft-only responses — a person sends, always
Open

Strategy & intelligence

  • Personal Operator grounded in the local operating record
  • Strategy council with independent specialists, red team, and judge
  • Operating memory that tracks sources, freshness, and contradictions
Open

Growth

  • Revenue targets back-solved into daily inputs
  • Six owned workflow blueprints, each with a human gate and closure receipt
  • Campaigns and experiments measured against real records
Open

Automation & authority

  • Visible loops: trigger, policy, gate, execution, readback, receipt
  • Version-bound Human Gates — editing a record invalidates stale approval
  • Bounded retries, provider readback, and rollback
Open

Data & recovery

  • Whole-workspace JSON, CSV, and SQLite export — leave anytime
  • Password-encrypted recovery files and rotating local backups
  • Row-level migration previews with repeat-import deduplication
Open

Team authority

  • Up to five named people with separate passphrases and roles
  • Native command enforcement — the browser UI is not the boundary
  • Durable allow/deny access receipts
Open

Mobile companion

  • QR pairing with desktop approval for every device
  • End-to-end encrypted projections through a content-blind relay
  • Review-only field capture — phones carry intent, never authority
Open

The architecture underneath

Seven layers, one rule: completion comes from reconciled state, not a model's opinion. The interface is built like an operator's instrument — exact, fast, calm under pressure, and visibly owned by a human.

Native shell
Windows desktop, tray, and native commands — the browser UI is not the permission boundary
Durable spine
Local SQLite with foreign keys, exact cents, events, receipts, backups, and migrations
Canonical graph
One identity chain across people, work, schedule, conversations, documents, and money
Intelligence routes
OpenRouter, loopback Ollama, or direct BYOK — no silent cross-provider fallback
Integration rail
Loopback signal API and signed webhooks with replay safety and a review inbox
Human Gates
Named, version-bound approval before money, commitments, or anything irreversible
Portability
Export, encrypted recovery, and mapping profiles designed in, not bolted on

Two more machines, their own doors.

The research fund with public receipts, and the live browser instruments.

Interactive system field
Open the complete systems atlas

Navigate the connected array of original systems, repositories, artifacts, and case files.

Expand Minimize

An arsenal of systems. One refusal to think small.

Explore an arsenal of public intelligence, simulation, infrastructure, security, and creative systems. Every node opens into real source—not theater.

DRAG FIELD · SELECT NODE · ARROW KEYS

Conceptual relationships, not runtime dependencies. Full-force presentation; human-controlled navigation, motion, and sound.

Development horizon
See what the portfolio is building toward

Explore the next generation of games, autonomous systems, forecasting machinery, and software-native operating structures.

Expand Minimize

We are building past the current map.

An upcoming array of original worlds and governed autonomous intelligence. The ambition is enormous. The release standard is harder: capability earns its name only after the evidence does.

VIDEO GAMES / INTERACTIVE WORLDS

Worlds that think back.

Original games where systems collide, characters adapt, and the world remembers what the player did. Built as living experiences—not reskins, content sludge, or another disposable loop.

  1. World
  2. System
  3. Player
  4. Consequence
UPCOMING / PLAYABLE SYSTEMS IN DEVELOPMENT
RECURSIVE PREDICTION INTELLIGENCE

An agent that improves how it knows.

A recursively self-improving autonomous prediction-market agent designed to form forecasts, run sealed experiments, learn from settlement, and promote only the methods that survive held-out evidence.

  1. Forecast
  2. Test
  3. Settle
  4. Improve
UPCOMING RESEARCH / PAPER-FIRST / EVIDENCE-GATED
AUTONOMOUS BROKERS / MARKET OPERATORS

Software-native hedge funds and prop firms.

Our north star: autonomous stock and options operators that research, decide, execute, reconcile, and learn across the complete market loop—inside explicit account mandates, hard risk limits, and an inspectable authority chain.

  1. Research
  2. Decide
  3. Execute
  4. Reconcile
UPCOMING VISION / LIVE TRADING REQUIRES SEPARATE AUTHORITY
AMBITION WITHOUT PRETENSEEVIDENCE BEFORE RELEASEAUTHORITY BEFORE ACTION