Skip to content
@ObtuseAI

Obtuse AI

Advanced systems studio. We build what others won't attempt—original architecture, evidence-backed machinery, human-owned authority.

Obtuse AI — We build what others won't attempt

ENTER THE MACHINE   ·   OPEN THE ARSENAL   ·   ARCHITECT YOUR ADVANTAGE   ·   CHALLENGE US

Impossible is usually an architecture problem.

Most teams retreat when the problem becomes tangled, high-consequence, or genuinely new. That is where we start.

Obtuse AI engineers the technically brutal, strategically important, supposedly impossible systems—the kind that demand original architecture, ruthless clarity, serious control, and proof strong enough to survive inspection.

We build intelligence engines, simulations, creative production systems, security research, and operator-owned infrastructure. Different machines. One operating doctrine:

Original by design. Evidence or it didn't happen. Humans hold the keys.

This is real source, real releases, and explicit authority—not pitch-deck fiction dressed as a product.

How impossible gets engineered

Attack Result
01 · Intent Strip away polite ambiguity until the objective, stakes, owner, and definition of victory are impossible to misunderstand. A brief strong enough to survive opposition.
02 · Architecture Turn interfaces, intelligence, authority, evidence, data movement, and failure paths into one deliberate machine. A system map with no hidden power and no weak seams.
03 · Construction Drive the critical path from raw input to controlled outcome. No fake doors. No hand-waved core. Software that performs the hard job end to end.
04 · Evidence Force every major claim through tests, receipts, provenance, replay, and adversarial review. An evidence trail that outlives the sales pitch.
05 · Authority Give the machine speed and leverage without letting it invent permission. Maximum capability. Consequential control stays human-owned.

The arsenal

An expanding array of public systems proving we mean it. Open the source, inspect the boundary, run the release, and see how far each machine actually goes.

Machine What it attacks Public boundary
Rampage
Case file · Live experience
Turns scattered local compute into one governed intelligence fabric with signed leases, encrypted artifacts, and operator-owned power. Local-first infrastructure · Proposal-only autonomous intelligence
DumbMoney
Case file · Live experience
Deploys bounded market specialists, then forces every conclusion through a deterministic evidence chain. Research system · Capital authority stays human
Dummy
Case file · Live experience
Forecasts crypto and sports with point-in-time evidence, calibration, paper twins, settlement learning, and exact replay. Public-source research · Paper-first · Human-gated authority
Jester
Case file · Live experience
Converts strategic hypotheses into deterministic causal worlds for market simulation, option economics, and scientific discovery. Simulation and research · Assumptions stay inspectable
Blunder
Case file
Unleashes agent engineering through isolated missions, deterministic validation, and receipts welded to every promotion decision. Operator owns integration
Dimwit
Case file
Commands Unreal, Blender, and multi-DCC production through bounded execution and independently checked artifacts. Human review required for creative claims
Doofus
Case file
Drives invention through multi-model debate and MAP-Elites search, then makes every candidate earn promotion on held-out evidence. Evidence-gated promotion · Operator owns project effects
Nimrod
Case file
Pushes cyber-defense research through constitutional limits, replayable attack scenarios, and independent verification. Research and simulation · No target or action authority
Waterboy
Case file
Drives multi-league sports simulation, auditable projections, hypothetical matchups, and deterministic replay. Simulation only · No wagering authority
Dopey
Case file
Interrogates equities and options with paper-only workflows, visible provenance, adversarial validation, and governed risk. Paper-only research · No brokerage authority

One doctrine. Multiple machines.

The repositories are separate products—not a version farm and not disconnected experiments. Each attacks a distinct problem while sharing an evidence and authority architecture.

flowchart LR
    IN["Intent<br/>Name the real problem"] --> AR["Architecture<br/>Design the advantage"]
    AR --> CO["Construction<br/>Build the critical path"]
    CO --> EV["Evidence<br/>Tests · receipts · replay"]
    EV --> AU["Authority<br/>Human-owned release"]

    EV -. "proves" .-> CO
    AU -. "bounds" .-> AR
Loading

Claims backed by machinery

We keep four questions separate because collapsing them is how serious systems turn into marketing theater:

  • Capability: What can the machine demonstrably do?
  • Evidence: What tests, artifacts, provenance, or replay support the claim?
  • Readiness: In what environment has it actually been validated?
  • Authority: Who can approve deployment, integration, capital, targets, or other consequential action?

Blocked, failed, inconclusive, and no-champion outcomes remain evidence. A green check is an engineering signal—not automatic proof of security certification, profitability, external validation, or live-operational readiness.

The historical public portfolio evidence snapshot is preserved as a dated audit artifact. The live arsenal above is the current navigation surface; each repository remains the authority for its own release, maturity, and operating boundary.

Bring us the problem everyone else keeps avoiding.

The tangled one. The high-stakes one. The thing with no template and no polite answer. We will tell you exactly what it takes—and then build it.

EXPLORE OBTUSEAI.NET   ·   BACK THE LAB   ·   PUT US ON IT

Advanced systems studio · Engineered in Kansas City, Missouri

Pinned Loading

  1. dimwit dimwit Public

    Evidence-native orchestration for Unreal, Blender, and multi-DCC production—bounded execution, receipts, and human review.

    Python 1

  2. nimrod nimrod Public

    Constitutional cyber-defense research for replayable simulation, cross-implementation verification, and fail-closed authority.

    Python 1

Repositories

Showing 10 of 11 repositories

People

This organization has no public members. You must be a member to see who’s a part of this organization.

Top languages

Loading…

Most used topics

Loading…