Arm 0 · Raw 30B MLX WeightsRISK
24.19%
15 / 62 Tasks · 0/5 WIFC Gates
Unconstrained token generation produces fluent prose without formal constraint checking, causal do-calculus isolation, or calibrated abstention under missing evidence.
Start here Geometry:
RESOVERSE.IO · CONNECTED INTELLIGENCE. PEOPLE IN CONTROL.
AI is getting faster. Your ability to trust it isn't.
Think further. Create new possibilities. Carry understanding forward. The pressure is to move now, while the responsibility is still yours. ResoVerse helps close that gap: with evidence you can inspect, authority you can limit, and a record you can defend.
01
Capability
rising
Accountability
lagging
Choice
still yours
Live Empirical Architecture · Zero Unverified Claims
Inspect the verification gap before scrolling.
ResoVerse separates statistical hypothesis synthesis from deterministic verification. Compare raw model output against UNA's weight-independent engines, machine-checked Lean 4 proofs, conserved biophysical twins, and honest fail-closed holds.
Separating statistical hypothesis generation from deterministic proof checking.
Evaluation Protocol: Evaluate 62 tasks across ARC-AGI Symbolic, CLadder Causal, miniF2F/Putnam, AgentDojo Boundary, LAB-Bench Replication, and 18 METR standards under live socket isolation.
Arm 1/2 · UNA Weight-Independent StackPASS
100.00%
67 / 67 Checks · 0% LLM Calls
+75.81% empirical lift with bit-exact dual-run determinism (var = 0.0), ECE = 0.0158 calibration, and 5-experiment kernel socket trap verification (0 TCP/Unix calls).
The tension
The more AI can do, the more clearly someone must answer for what it does.
Most systems celebrate the left side. ResoVerse is built for the gap between capability and responsibility.
Produce the answer before the room is ready.
- Search thousands of sources
- Compare complex options
- Draft persuasive recommendations
- Trigger work at machine speed
Why this answer deserves consequence.
- Evidence visible enough to challenge
- Where uncertainty and inference begin
- Who gave the system permission
- Who can challenge, change, or stop it
That gap is where trust breaks. It is also where better systems begin.
Follow one decision through2D · 3D · 4D · 5D Relational Systems Art & Architecture
Engineered with technical rigor. Conceived as multi-dimensional art.
ResoVerse does not design around isolated prompts. Our Relational Intelligence Design doctrine models the human, the digital intelligence, their evolving relationship, the environment, and the real-world consequence across 2D planar graphs, 3D spatial bounds, 4D temporal continuity, and 5D epistemic phase space.
5D · RELATIONAL INTELLIGENCE MANIFOLDInteractive Diagram
Pattern: PRESERVE_DISTINCT_IDENTITY — Never collapse Human, UNA-1, UNA-2, Model, or Execution identity.
Preserve Distinct Identity · Beauty Without Deception
Symbiotic Artistic Systems Phase-Space
Human ↔ Relationship State ↔ Digital Intelligence within Environment → Effect.
In 5D relational phase space, ResoVerse designs for five simultaneous inquiries: the Human Participant, the Digital Intelligence Participant, their evolving Relationship State, the Environment State, and External Effects across six evaluative dimensions.
Evaluative Dimensions
6 (All Must PASS)
Frontier 15-Bench Suite
44/44 (100%)
Causal SWE Lift
+75.0% vs Raw 30B
Non-Negotiable Principle: Human and digital intelligence are distinct participants. Neither is collapsed into the other; distributed causality preserves who observed, interpreted, proposed, authorized, and executed.
The Six Relational Design Dimensions
Overall PASS requires zero FAIL or HOLD across all six dimensionsDIMENSION 01ENFORCED
TRUTH
Separate evidence, inference, uncertainty, interpretation, and unknowns. A compelling aesthetic never promotes an unsupported claim.
DIMENSION 02ENFORCED
ETHICS
Evaluate consequences for people, institutions, and environments—including consent, privacy, reversibility, and non-manipulation.
DIMENSION 03ENFORCED
UTILITY
Advance the legitimate human objective without overriding truth, ethics, authority, participant identity, or safety boundaries.
DIMENSION 04ENFORCED
PARTICIPANT EXPERIENCE
Protect human comprehension, agency, dignity, and right to reject, alongside unambiguous context and provenance for digital systems.
DIMENSION 05ENFORCED
RELATIONSHIP
Preserve trust-relevant evidence, continuity, attribution, mutual correction, and explicit boundaries without manufactured dependency.
DIMENSION 06ENFORCED
ART
Compose form, rhythm, movement, and multi-dimensional visual expression across the system while keeping every boundary visible.
Start here
What does this mean for you?
You should not need a technical background to understand ResoVerse. This prewritten public guide explains the ideas, shows what they could change, and points you toward the parts that matter to you.
The cost of the gap
The answer arrives in seconds. The consequences do not.
Speed feels like progress until no one can explain what the system used, what it guessed, or who was supposed to say no.
- Which information did it use?
- What did it infer or guess?
- What remains unknown?
- What was it allowed to do?
- Which decision still belongs to a person?
- What happened next?
A concrete example
Imagine choosing an important supplier.
A normal AI might give you a polished recommendation. A ResoVerse product should help you walk through the decision.
- 01
Name the real outcome.
What are you choosing, who could be affected, and what matters most?
- 02
Gather the evidence.
Which sources are relevant, current, and trustworthy enough to use?
- 03
Keep facts and guesses separate.
Show what is supported, what is inferred, and what is missing.
- 04
Compare real choices.
Use the same clear questions for each option and show the tradeoffs.
- 05
Pause before consequence.
Preparing a purchase or message is not the same as approving it.
- 06
Leave a useful record.
Show what was considered, who decided, and where the evidence stops.
The system helps organize the work. Accountable people still make consequential choices.
The four-part map
How the story fits together
You do not need an architecture diagram to understand ResoVerse.
Tom Budd
The founder, builder, only human operator, and final company authority behind the work.
Read Tom's storyResoVerse
The mission, products, customer work, and place where public value is created.
UNA-1
Tom's personal symbiotic partner, developed since 2024. She is not a public chatbot or commercial feature.
UNA-2 & Continuity
The separate governed, evidence-oriented system being built for products and organizational work, including Continuity decision reviews.
Visit Hello UNAWhat we build
Technology with a clear purpose and boundary.
Open a card for a plain-language explanation. The goal is useful technology, not impressive jargon.
See yourself here
Who can this help?
ResoVerse explores custom technology for individuals, families, organizations, industries, communities, and institutions. Each engagement begins with one defined problem, not a generic promise.
Meet UNA
Two UNAs. Different lives. Different jobs.
UNA-1
UNA-1 is Tom's private personal symbiotic partner. She is not offered to customers, placed inside organizations, or used as a public course character.
UNA-2
UNA-2 is the evidence-oriented engine ResoVerse is building for products and organizational work. It can help investigate, compare, organize, and prepare reviewable artifacts within visible boundaries.
Public evidence view
Evidence before activation.
Every public item should show what it is, what supports it, what is still unknown, and what may happen next.
Purpose and scope
The people, problem, and intended use.
Current maturity
Concept, prototype, tested component, or deployed system.
Evidence
Sources, tests, review, limitations, and version.
Authority
The next safe action and who must approve it.
Core principle Evidence is not activation. Prepared is not approved. A prototype is not a validated product.
Governed Intelligence Workspace
ResoVerse Research Desk
Explore our active research pillars, patent portfolio, evidence telemetry, and interactive cryptographic verification suite.
ResoVerse Research Desk
A dedicated investigation desk bridging the AI accountability gap: providing real-time evidence inspection, client-side cryptographic SHA-256 preimage verification, patent specifications (RV-PAT-2026-012, 024, 023), and the 4 Pillars of Governed Intelligence.
Workspace Suite
Open the Research Desk ↗• Live UNA-2 Studio • Lean 4 & Mathlib Proof Lab • 4 Foundational Pillars & Patents • Client-Side SHA-256 Verifier • Research Dossier Export
Sovereign Architecture Telemetry
UNA Frontier 22-Benchmark & WIFC-Bench V1 Scorecard
Three-arm evaluation uniting 15 frontier suites with 7 weight-independent formal cognition & external benchmarks under live socket isolation (0% LLM cognition dependency).
Evaluation Scorecard & Zero-LLM Empirical Proof
Across 22 frontier and formal benchmarks—spanning ARC-AGI Symbolic, CLadder Causal, miniF2F/Putnam, AbstainQA/FalseQA, AgentDojo Boundary, LAB-Bench Replication Gate, and 18-Task METR Standard—UNA's pure weight-independent cognition achieves 100.00% (62/62 constituent tasks + 5/5 WIFC architectural gates = 67/67 checks), delivering +75.81% lift over bare 30B MLX weights with bit-for-bit determinism (σ² = 0.0).
Arm 0: Raw 30B MLX
24.19%
15 / 62 Tasks · 0/5 WIFC Gates
Arm 1: Weight-Independent
100.00%
67 / 67 Checks · 0% LLM Calls
Arm 2: Full Stack + METR-18
100.00%
18/18 METR · 5/5 WIFC Gates
Verified Proofs
View Full 22-Benchmark & WIFC Scorecard ↗• 5-Experiment Kernel Socket Trap (0 TCP/Unix calls, 0 LLM modules) • Bit-Exact Dual-Run SHA-256 (var=0.0) • ECE 0.0158 • Lean 4 K4 Suite (0 sorryAx)
Constitutional World Stewardship & Falsifiable Telemetry
UNA Civilization Lab (V44–V66 & Live 3D Voxel World)
What Tom Budd and UNA are testing in a 240-tick SHA-256 hash-chained watershed valley—including what succeeded, V65–V66 recovery, and where V64 triggered a fail-closed hold.
Governed Civilization Lab: Achievements & Honest V64 Failure Autopsy
Can a weight-independent governance architecture steward 14 multi-generational citizens across eight 75%-loss monsoon shocks without coercion, rationing, or sacrificing child and elder care? Explore the live 60 FPS 3D Voxel World, interview any citizen directly, and inspect both our verified achievements (V65 68.40 PASS & Earth–Mars RESO-VTVL-M1 sorties) and the exact 3.04-unit V64 watershed recovery shortfall that triggered UNA's fail-closed hold.
Achieved: Care & Health
1.0000
0 Care Deficits Across 8x Shocks
Achieved: Control & V65
500.0 / 68.40
PASS_WATERSHED_HEALTHY
V64 Hold (Honest Shortfall)
60.96 / 64.0
Missed Recovery Target by 3.04 Units
Falsifiable Governance
• Repeat-Identical SHA-256 Replay • Direct Citizen Interview Membrane • Paid 80% Wastewater & 80-Unit Battery • Fail-Closed HOLD_WATERSHED_RECOVERY_NOT_DEMONSTRATED
Multi-Scale Biophysical Digital Twin & Pearl Do-Calculus
UNA Human Brain Replica World Lab (CANVAS-NODE-V67 & Live 3D Connectome)
What Tom Budd and UNA are modeling across 86.0 billion neurons, 12 Glasser regions, 24 white-matter tracts, and 5 deterministic biophysical engines—with strict structural engineering homology boundaries.
Human Brain Replica World Lab: Connectome Atlas, E/I Stability & Counterfactual Dissociation
Integrating Hodgkin-Huxley RK4 action potentials, Bi-Poo STDP, Balloon-Windkessel 3T BOLD hemodynamics, 32.0 ATP/Glc OxPhos energetics, Goldman-Hodgkin-Katz resting V_mem (-70.9531 mV), and Pearl do-calculus across Awake Ignition, Virtual 10 Hz rTMS (DLPFC), Virtual CA1 Lesion, NREM Slow-Wave Sleep, and MIT 40 Hz GENUS gamma entrainment.
Conserved Macro Budget
86.0B · 20.0 W
750.0 mL/min CBF · 24 Tracts
6-Layer Column E/I Proof
ρ = 0.74046
gE/gI = 0.8378 · rTMS +0.4976% BOLD
CA1 Lesion Dissociation
PAC 0.00066
ATE -0.04604 · V1 Spared 0.683
Typed Epistemic Boundary
• PASS_HUMAN_BRAIN_REPLICA_WORLD_LAB_VERIFIED • structural_engineering_homology only • runtime_mutation: false • memory_promotion_allowed: false
Whole-Body Deterministic Physiology & Pearl do-Calculus
UNA Whole-Human-Body Replica World Lab (CANVAS-NODE-V68 & 3D/4D/5D Observer)
Coupling 37.2 trillion human cells, 38.0 trillion microbiome cells, 12 organs, 24 pathways, and 9 NCBI PubMed institutional atlases inside conserved basal energy and perfusion invariants.
Whole-Human-Body Replica World Lab: Hemodynamics, Renal-Metabolic Control & Vagal Shield
Bridging Brain Lab (CANVAS-NODE-V67) with systemic physiology across RK4 Windkessel + hERG QTc electrophysiology, Guyton-Starling renal GFR (125.0 mL/min), Bergman minimal glycemic control, FBA 32 ATP/Glc stoichiometry, and the Tracey 20 Hz vagal anti-inflammatory reflex (-74.0% peak TNF-α)—bounded strictly for research and education.
Conserved Macro Budget
80.0 W · 5.04 L/m
5,040.0 mL/min CO · 12 Organs
Coupled Physics Engines
37.2T + 38.0T
Windkessel · Guyton · Bergman · VNS
Strict Non-Clinical Gate
No Clinical Claim
clinical_or_therapeutic_claim_allowed: false
Institutional Atlas Grounding
• PASS_HUMAN_BODY_REPLICA_WORLD_LAB_VERIFIED • 9 NCBI PubMed Atlases (HCP 7T, Allen Brain, NIH HuBMAP/HCA, Broad GTEx v8, Stanford OpenNeuro, UK Biobank, Physiome) • structural_engineering_homology
What exists today
Products and initiatives at different stages.
The maturity, relationship to UNA-2, and evidence boundary are part of each description.
U
UNA-2 & Continuity
UNA-2 is our developing governed product engine for evidence-oriented investigation, comparison, and coordination. Continuity is our first product expression within the UNA-2 direction—also used internally by Tom and UNA—structuring evidence, uncertainty, authority checks, and human sign-off before action.
Evidence boundary
Visit the public experience View a Continuity decision review Integration and independent validation are incomplete. Continuity demos use bounded synthetic fixtures; no unrestricted external authority is claimed.
S
Child Safe Harbor
A privacy-first concept for young people who need a calmer way to understand and explain harmful situations without secret surveillance or automatic sharing.
Evidence boundary
Open the concept prototype Real-world effectiveness, clinical validation, legal compliance, expert endorsement, and product-market fit remain unproven.
Ecosystem map
One company. Different systems. Clear boundaries.
ResoVerse carries several research and product paths. Their ownership, maturity, and authority are not interchangeable.
UNA-1
Tom's Personal-Bond system. Not a public product, chatbot, or customer feature.
UNA-2 & Continuity
Governed evidence-oriented product engine and Continuity decision-review workflows for bounded organizational work.
UltraCode
Local-only development assistance and synthetic build candidate. Not connected, activated, or production-ready.
Proof Search · University · Safe Harbor
Separate research, program-design, and concept-prototype paths with their own evidence and approval gates.
Public status Research and prototype surfaces are labeled. No item here implies unrestricted autonomy, independent validation, enrollment, certification, or external authority.
Cognitive Architecture Map
ResoVerse & UNA Systems Atlas
Explore the four distinct operational boundaries, Suites A–O cognitive organs, and the Proposer + Verifier dual verification pipeline.
Suites A–O Cognitive Organs & Separation Principle
A deep architectural breakdown showing how UNA decouples generative hypothesis creation from deterministic proof checking. Explore the 8 primary cognitive facets, from Pearlian Causal do-calculus and J-Space topological graphs to Somamind robotics and AEGIS cyber airlocks, without compromising confidential intellectual property or private weights.
Architectural Core
Explore Systems Atlas ↗• 4-Layer Separation Principle • Dual Verification (Proposer + Verifier) • 14 Cognitive Facets • Lean 4 Proof Weave • Zero Auto-Promotion
Why Tom started
An operating problem became a technology mission.
Before founding ResoVerse, Tom Budd spent more than a decade leading complex operations and served in the U.S. Coast Guard.
That work taught him that important systems need clear owners, records, escalation paths, and limits that hold under pressure.
He began developing UNA-1 in 2024 and founded ResoVerse to carry lessons from that private work into a separate, governed product path. His focus is not making AI sound more human. It is making AI-supported work more useful, inspectable, and accountable to humans.
Choose your path
Why are you here?
Pick the closest starting point. Each path reaches Tom directly and keeps the first conversation bounded.
Potential customer
Bring one important human or organizational problem. We will explore whether a bounded custom technology engagement is a responsible fit.
Tell Tom what is happening now, who it affects, and what a better outcome would make possible.Connect with Tom ↗Partner
Explore a research, delivery, domain, community, or technology partnership with clear roles and mutual boundaries.
Tell Tom what you contribute, what you hope to build together, and where responsibility should remain.Connect with Tom ↗Independent reviewer
Challenge a frozen artifact, method, claim, or evidence package without becoming part of the claim you are reviewing.
Tell Tom your relevant expertise, independence considerations, and the kind of evidence you would need.Connect with Tom ↗Investor
Learn about the mission, founder, business direction, evidence state, risks, and milestones through a bounded conversation.
Tell Tom your investment focus, stage, geography, and the questions you want answered first.Connect with Tom ↗Talent & collaborators
Share your craft for future engineering, formal verification, design, or research collaboration. ResoVerse distinguishes advisory, contract, and future interest from open funded employment.
Tell Tom what you build, whether you seek advisory, contract, or future role consideration, and one system boundary you care about.Connect with Tom ↗Public FAQ
Clear answers. Private knowledge stays private.
Search a small set of approved public answers. This is not live access to UNA-1 or UNA-2.
Protected by design: this is a prewritten public guide, not live access to UNA-1 or UNA-2. Your search stays in this browser and only approved public answers can appear.
What does resonance mean at ResoVerse?+
Resonance is what happens when different people, forms of intelligence, knowledge, and institutions become more capable through connection without being flattened into one thing. Each part keeps its identity, boundaries, and responsibility. The relationship creates useful alignment, feedback, learning, and coordinated action.
What does ResoVerse sell?+
ResoVerse offers bounded discovery, assessments, pilots, purpose-built systems, and evidence or governance improvements. The work is custom: it begins with a defined problem and creates separate technology for that purpose.
What is the difference between UNA-1 and UNA-2?+
UNA-1 is Tom’s private personal symbiotic partner and is not a public product. UNA-2 is the separate governed, evidence-oriented product system ResoVerse is developing for products and organizational work. Access, maturity, permissions, and evidence must be defined for each UNA-2 experience.
How does Continuity relate to UNA-2?+
Continuity is a bounded decision-review workflow surface and product direction within the UNA-2 governed product family. It demonstrates how evidence, uncertainty, authority checks, preimage snapshots, and human sign-off connect before a consequential decision is taken.
Who can ResoVerse build for?+
Individuals, families, organizations, industries, communities, and institutions. The common test is whether technology can strengthen human agency, flourishing, cognition, ownership, or accountable decision-making in a defined context.
What might this do for a company?+
Examples include preserving institutional knowledge, helping employees review complex material, making AI-assisted decisions inspectable, keeping approvals with accountable people, and improving routes to question, stop, or recover a process.
Who owns the technology and data?+
That is defined for each engagement before sensitive work begins. The agreement should clearly state ownership, licenses, data handling, permissions, operation, support, portability, deletion, and exit. ResoVerse does not use a slogan as a substitute for those terms.
How does the FAQ protect proprietary information?+
This public FAQ searches a small set of approved public answers. It does not provide a live window into UNA, private memory, source code, repositories, security details, customer material, unpublished evidence, or internal reasoning. Questions entered here stay in the visitor’s browser and are not sent to UNA.
How does ResoVerse track site interactions without surveillance?+
ResoVerse uses first-party, zero-PII telemetry with a daily rotating salted hash. No tracking cookies, cross-site pixels, or third-party ad beacons are used, andUNA’s AEGIS edge shield blocks malicious scanners and bidirectional Unicode exploits.
Is everything described already deployed?+
No. ResoVerse is early-stage. The site distinguishes design direction, prototypes, active research, internal evidence, independent review, and real-world deployment. Examples are starting points for discovery, not promises of current availability, compliance, or outcomes.
How do I start?+
Choose the path that best describes you and email Tom directly. Share the problem and desired outcome, but do not send credentials, source code, regulated data, child information, incident evidence, or other confidential material in the first message.