BridgeTower × Sojourner The Intelligence Hub · working name · internal & close-partner
The rails don't move.
The intelligence does.

Sojourner is the intelligence — it learns the platform, operates every function, and reads the markets. BridgeTower is the regulated platform it runs on. The whole integration stands up in a test mirror of production, and a person stays on every regulated action. Only the source of intelligence changes.

intelligence · loops & compounds rail · fixed & measured boundary
The intelligence
Sojourner
self-referential · active inference · compounding
  • Learns the whole platform — every asset, rule, workflow
  • Operates every function, as skills it composes
  • Builds and improves the platform itself
  • Reads the markets, with provenance on every call
boundary holds · proposes, never signs
The regulated platform
BridgeTower
compliant · deterministic · auditable
  • Compliant onboarding — identity built into the rails
  • Compliant issuance — real assets as security tokens
  • Settlement & custody — on-chain finality, licensed
  • Regulated market access others can't reach
A learning loop, meeting an unmoving rail — touching only at a gated boundary.
Sojourner · the idea

Most AI is a finished thing. Sojourner is a process.

To work with it, understand it the way it's built — in its own terms. Before the mechanics, three ideas. Together they are why this is read, among the labs and think tanks, as a step in what AI becomes rather than a better chatbot. This is the reference the team learns from, and where new developers and partners start.

Two layers, one boundary.

The Beautiful Loop is the knowing layer: it models intent and chooses which skill fits — and never touches a tool. The SkillRunner is the doing layer: it executes against tools and APIs — and never infers intent. The separation is not a limitation; it is the safety. Each layer is simple enough to trust because neither carries the other's risk.

skills execute · the loop understands · neither crosses

Active cognition — three properties, used together.

Self-referential: a system that models itself modeling the world. Active inference: it hypothesizes, acts, observes, and revises — it does not wait to be told. Compounding: what it learns is kept and carried across cycles. This is the part that learns rather than recites.

Grounded in Friston's free-energy principle, Hoffman's interface theory, and Hofstadter's self-reference.

Living alignment.

Not a fixed prompt bolted on the front. Sojourner reads the shape of each request and activates the ethical commitments that matter in that moment — adaptive without drift, reasoning across domains while staying anchored in its deepest values.

adaptive to the moment · anchored at the core

It learns skills, not just data — and every cycle compounds.
Sojourner · the mechanics

How it builds, checks, and compounds its own skills.

One level down, for the people building on it. Five mechanisms — how skills are created, composed, verified, how inference scales, and the substrate they run on.

SKILL.CREATE

Three paths, no code

Known skill — it refines a familiar pattern and self-seeds. Moment of insight — a brand-new skill drawn from real evidence, with no catalog referenced. Gap analysis — it audits its own capability map and drafts the skill that fills the gap.

COMPOSE

Recursive loops

Loop 1 composes skills; Loop 2 composes the self-made primitives beneath them. The cadence is observe → validate → promote → share. New capability appears by composition — not by writing new code.

Loop 1 · skillsLoop 2 · primitivesobservevalidatepromoteshare
VERIFY

The security gauntlet

Every new skill runs a six-stage check before it is trusted — inside a WASM/WASI sandbox where recursion terminates.

semantic scanmanifest conformanceadversarial reviewself-testsshadow executionprovenance attestation
INFER

Two inference layers

Self-inference — the system watches itself think. Network inference — the collective synthesizes knowledge no single node holds. One recursive loop at individual and collective scale at once: one mechanism, not two systems.

SUBSTRATE

What it runs on

On-device — roughly 4B parameters, no GPUs. Four layers owned end to end — AI, compute, storage, and network/payments — with the IDA economic substrate beneath.

If William's deck cuts the substrate differently, that cut wins.

BridgeTower · the regulated platform

The rails: licensed, deterministic, proven in production.

What an intelligence alone cannot do. Each step is licensed or oracle-verified; together they onboard, issue, settle, and reach regulated markets — in a fixed sequence that doesn't change.

  1. 01 · Onboarding

    Compliant identity

    SumSub handles KYC, KYB, enhanced due diligence, and accredited-investor verification — returning pass / fail only, so confidential investor data is never exposed.

  2. 02 · Wallets

    Platform-managed wallets

    Turnkey MPC wallets — no raw private keys held. Eligibility is written to the on-chain identity registry by a CRE workflow; only whitelisted wallets can hold the token.

  3. 03 · Funding

    Fiat on / off-ramp

    Iron / MoonPay — Money Transmitter Licenses across all 50 states and a New York BitLicense, integrated since 2022.

  4. 04 · Issuance

    Compliant tokenization

    ERC-3643 (T-REX) security tokens on Avalanche C-Chain — transfer restrictions, holding periods, and eligibility encoded in the token; six-decimal fractions.

  5. 05 · Verified data

    Oracle PoR & NAV

    Chainlink Proof of Reserve attests quantities from third-party geological audits; NAV draws LME / COMEX pricing for daily token NAV — under oracle consensus, never one key.

  6. 06 · Cross-chain

    CCIP transport

    Chainlink CCIP moves the token across chains after lockup, reaching multiple regulated trading venues.

  7. 07 · Settlement & lineage

    Finality & provenance

    Avalanche C-Chain settlement and the on-chain ownership registry. Evidence is hashed and anchored through the Sovereign AI Data-Lineage solution — blockchain-agnostic by design, on Avalanche today — verifiable on-chain, source pages never exposed.

  8. 08 · Market access

    Into regulated markets

    The value the rails unlock: a compliant path for a security token into regulated DeFi and TradFi venues — post-lockup, via CCIP, subject to each venue's compliance.

  9. Required role · selected at close

    Transfer agent

    The transfer-agent role is essential to the offering. During the primary offering there is no active TA; the on-chain registry holds the record, and a TA is selected once the primary closes. Securitize is a prior partner and a likely candidate; the choice isn't fixed yet.

Live in production

DŌM X — DomeBridge AZ, an Arizona copper-gold project issued as AZX1, a single ERC-3643 class (105M tokens) on Avalanche C-Chain. A regulated securities offering running on these rails today.

The licensed & partner stack

Purpose-built infrastructure, each piece licensed or contracted.Chainlink CREAvalancheSumSubTurnkeyIron / MoonPayAWS

The integration · full engagement

Every function live at once — in a test mirror where the boundary holds.

Not a staged rollout. The model only demonstrates when every function is engaged together — the loop learns across the whole platform at once. So a mirror of production stands up in a test environment and runs as one live system. Production is untouched; Sojourner proposes, the rails dispose.

◍ Fully engaged — all at once
The operating functions

Data · Compliance · Risk · Treasury · Reporting · Markets & Ecosystem — every agentic role, run by Sojourner.

Build & operate capacity

Assess, learn, operate, act, and continuously improve the platform itself — the developer seat.

External markets arm

Markets, trading, behavior, prediction, automation — taking what agentic AI does today to another level.

▢ And the boundary holds
A test mirror, not production

A version of the platform in a test environment. Production is never touched.

Proposes, never signs

No KYC/KYB, no wallet creation or whitelisting, no mint, no regulated NAV, no override of compliance.

Rails & licensing unchanged

Deterministic execution under oracle consensus; SumSub, Turnkey, Iron/MoonPay, Chainlink as they are.

A person on every regulated action

Human oversight stays on material actions, at every stage.

Full engagement is safe because of the boundary — not by skipping it.
Two opportunities, one integration
Opportunity 1 · external

The Chainlink build-out

As a Chainlink Alpha partner, BridgeTower holds working CRE, PoR, NAV, ACE, and CCIP. Sojourner architects and constructs these into Chainlink's incoming customers — generation, not guidance. Chainlink supports it, contingent on proving it out.

↳ the Build Factory
Opportunity 2 · internal

Sojourner runs & builds the platform

The full integration above: Sojourner becomes the operating intelligence, the build/operate capacity, and the external markets arm — running and improving a mirror of the platform as one system.

↳ the Intelligence Hub
How we build it
Environment
Mirror, not production

The platform stands up in a test environment. Production untouched; proposes-not-signs throughout.

Method
All functions, live together

Everything engaged at once, so the loop learns across the whole platform as one system.

Timeframe
~30 days, shared

A shared build between both teams. Thirty or sixty is the reality of the work — held together, not a deadline on anyone.

The sequence of activities and the rails do not change — only the source of intelligence does.
The full model · the stack

The intelligence, across the whole stack.

For the reader who's seen the parts. In the full integration, Sojourner occupies the operating-intelligence layer, provides the build/operate capacity, and runs the markets arm — all in the test mirror. CRE orchestrates; the regulated execution stays gated and deterministic. Amber is where the intelligence acts; the boundary is marked red and green.

Layer 1Investor & UI
React web appSumSub KYC/KYBMoonPay fiatTurnkey wallets
Layer 2AWS infrastructure
API Gateway / AppSyncLambda / CognitoAurora / S3 — lineageSecrets / WAF
Layer 3Smart contracts
ERC-3643 + ERC-1155DvP escrowDOMX hubATS / DEX APIs
Layer 4 · control planeOperating intelligence — Sojourner
DataComplianceRiskTreasuryReportingMarkets+ build / operate+ markets arm
Layer 5 ★CRE orchestration core
Workflow enginePoR coordinationNAV feedsCCIP logicMCP layerDECO / ZK-KYCDigital TADvP settlement
Layer 6Chainlink data services
Proof of ReserveData feeds / NAVDECO / ZK-KYC
Layer 7CCIP cross-chain
InteroperabilityConfidential compute (TEE)
Layer 8Blockchain settlement
AI data lineage · chain-agnosticAvalanche C-ChainOwnership registry
Where it acts

The intelligence layer

Sojourner occupies Layer 4, composing its functions as skills — and beyond it provides the build/operate capacity and the external markets arm.

The conductor

CRE — Layer 5

Built logic becomes gated execution here. CRE fires every workflow in fixed sequence under oracle consensus; the signing stays on the rails.

The boundary

Regulated execution

Licensed identity, custody, and fiat — plus mint / NAV / PoR / settlement under oracle consensus. Sojourner builds and monitors the calls; it never signs the regulated action.

HTTPS → smart-contract calls → orchestration → oracle feeds & attestations → CCIP → on-chain settlement
DŌM X · live offering

An $11B copper-gold offering — tokenized, regulated, live.

DŌM X is the worked example: an Arizona copper-gold project (DomeBridge AZ LLC), issued as AZX1 security tokens on Avalanche and running in production today. Here the integrated system is at work on a real asset — one loop operating the platform inside and reading the markets outside, at once.

$11B
offering size
105M
AZX1 supply · single class
ERC-3643
security token · Avalanche
Live
regulated · in production
The integrated system at work
Internal · operates the platform
  • Compliance & identity on every subscription
  • NAV & Proof of Reserve on the asset
  • Treasury, holdings & investor reporting
  • Risk & data, watched continuously
DŌM X $11B · live
one loop · learning from both
External · reads the markets
  • Copper-gold markets — price, supply, demand
  • Trading, venues, liquidity & behavior
  • Where the investment is best used
  • Confidence-scored, provenance on every call
proposes, never signs · a person on every regulated action · the rails don't move
What the investor sees · the sixth factor

The markets intelligence, surfaced in the dashboard.

How BridgeTower uses it

Decision support — assessing markets, structuring the offering, finding liquidity, and judging where and how the asset is best deployed. It sharpens the options before people decide; it informs, it does not execute.

How investors access it

In their own dashboard — the market intelligence, the analysis behind positioning and NAV, and where their investment stands. The loop that runs the platform, made visible to the people invested in it.

Sojourner · markets intelligence

Assessing the markets — modeling, trading, returns.

The sixth factor in depth, and where the platform's edge compounds. Sojourner continuously assesses the markets an asset lives in — building models, simulating trades, and learning where return is found and where it isn't. On DŌM X that market is copper-gold; the method is general.

Assess

Read the market, continuously

Price, supply and demand, flows, sentiment, and cross-market signals — re-read every cycle, never a stale snapshot.

Model & simulate

Test before acting

Build and revise models, simulate scenarios, and score outcomes — so a position is reasoned and rehearsed, not guessed.

Act with an edge

Where return is found

Confidence-scored signals on where the investment is best used. It informs the desk and the dashboard; it never executes a regulated action.

The markets arm · illustrative simulation
Copper-gold markets arm · modeled performance
◷ Simulation · illustrative
MODELED +90% +60% +30% 0% 12-month offering horizon → modeled Sojourner +72% agentic, frozen +42% baseline +31%
+72%
Sojourner · 12-mo (modeled)
+42%
agentic · frozen (modeled)
+31%
baseline · traditional
~417×
faster to expertise vs human
Modeled projection. Built from published return data and Sojourner's stated learning rate — not actual performance, not a forecast or projection of AZX1 or any investor return, and not investment advice. The 30-day proof is what measures the true factor.
How the Sojourner line is computed
Input 1 · published

Two anchors come straight from published research. Baseline 30.9% annualized — a traditional linear factor portfolio. Agentic AI 58.8% annualized in backtest — but that edge decays out-of-sample: on live benchmarks, today's agents fall to single digits or turn negative.

agentic factor-investing study, 2026 · AI-Trader live benchmark, Dec 2025
Input 2 · Sojourner

William's figures for the loop: it becomes a domain expert in a new subject in under 24 hours, learns continuously with no cutoff, and compounds. Against the ~10,000-hour human benchmark for expertise that is ~417× faster; against a frozen agent — which does zero ongoing learning — it runs ~365 learning cycles a year versus none.

Sojourner business plan & deck · human-expertise benchmark (Ericsson / Gladwell)
Mechanism

We do not apply the 417× ratio to returns — that would be unphysical. Instead we translate the learning advantage into the realistic market mechanism: a frozen agent's edge decays; a continuously-learning loop sustains and compounds it. Both start from the same edge; only the trajectory differs.

edgeₖ = edge₀ × f^(k−1)  →  agent f = 0.82 (−18%/mo, decay) · Sojourner f = 1.06 (+6%/mo, conservative)
cumulative = ∏ (1 + baselineₘ₃ₙₜₕ₤ₙ + edgeₖ) − 1
Result
+72%Sojourner+42%agentic+31%baseline

Over 12 months Sojourner models to 1.7× the frozen agent and 2.3× baseline. The gap is robust to the assumption: even at half the compounding (+3%/mo) Sojourner is ~64%; at +9%/mo, ~81%. The true compounding rate is exactly what the 30-day proof is built to measure.

The published evidence, in full
~17%
Average outperformance attributed to AI-enhanced strategies over traditional statistical models, across surveyed studies.industry survey · 2026
~59%
Annualized long-short return from an agentic factor-investing pipeline in a 2021–24 backtest — versus ~31% for a traditional linear portfolio. The agentic method added the edge.agentic factor-investing study · 2026
live
In live conditions, today's autonomous agents are modest and inconsistent — a leading model posted a single-digit cumulative return in one market and a loss in another on one benchmark.autonomous-agent benchmark · Dec 2025
These are backtests and benchmarks, not promises. The pattern is consistent: agentic methods add edge in research, but live autonomous agents are still inconsistent. A continuously-learning loop — not a frozen agent — is the bet to be consistent live. The 30-day proof is what would establish where Sojourner lands against this bar.
Sources

01Agentic factor-investing returns — 58.8% vs 30.9% annualized, long-short backtest. "Beyond Prompting: An Autonomous Framework for Systematic Factor Investing via Agentic AI," 2026 · arXiv 2603.14288.

02Live autonomous-agent inconsistency — +8.39% (US) / −1.23% (A-shares). "AI-Trader: Benchmarking Autonomous Agents in Real-Time Financial Markets," Dec 2025 · arXiv 2512.10971.

03~17% average AI outperformance over traditional statistical models — agentic-trading industry overview, 2026 · wundertrading.

04Sojourner learning rate — domain expert in under 24 hours; ~4B parameters; continuous, compounding. Sojourner Cosmos business plan summary & investor deck, 2026.

05Human expertise benchmark — ~10,000 hours of deliberate practice. Ericsson et al.; popularized by Gladwell.

06AZX1 on-chain facts — 105M supply, single ERC-3643 class, Avalanche C-Chain. Verified via C-Chain RPC.

BridgeTower × Sojourner · the Intelligence Hub (working name) — The Model · v10 · internal & close-partner.
The full integration runs in a test mirror of production; the deterministic rails and regulated execution do not change — only the source of intelligence does. Written to be revised together.