Running as Five Systems
The homepage table showed best-in-class systems at their best. This page shows the full picture - complex configurations where the problem is orders of magnitude worse, the adjudication exposure that doesn't appear in latency numbers, and the capital cost of running five operations as five systems.
The Full Comparison Table
The homepage table showed best-in-class traditional infrastructure and blockchain under optimal conditions - DTCC same-day settlement, Ethereum/Solana on-chain with no congestion. Even there, full economic finality is T+1 for both, because accounting sync remains manual or batch in both cases.
The table below shows the full picture. Optimal conditions appear at the top of each range. Complex instruments - trade finance, structured products, cross-border transactions - appear at the bottom. The Conduit column does not have a range. The upper bound is the same for any transaction at any complexity.
Scroll to see full comparison →
COMPARISON DIMENSION | 01 · LEGACY Traditional Systems | 02 · WEB3 Blockchain | 03 · CONDUIT OS Conduit OS |
|---|---|---|---|
01 Agreement Execution | 2 weeks - 6 months | Partial encoding only | CONDUIT OS RUNTIME one atomic event Agreements Authority Rights Settlement Accounting |
02 Authority Enforcement | Implied (org structure) or asserted (legal agreement) | Authentication + application-coded authorization | |
03 Rights Recognition | T+1 (US equities) / 30-60 days | ~400 ms - ~5 s (on-chain token) | |
04 Settlement | T+1 - Net-90 (trade finance) | ~400 ms (Solana) - ~12 s (Ethereum) | |
05 Accounting Sync | Hours-days (batch) / 5-10 day close | Real-time (on-chain ledger) | |
| Dispute Adjudication | 6 months - 5+ years | 6 months - 5+ years | Query the ledger |
| Time to Full Economic Finality | T+1 (payments) | Days-weeks (new) | Under 100ms |
Sources: World Commerce & Contracting (agreement execution); SEC T+1 rule May 2024; ICC Trade Register (trade finance); APQC enterprise benchmarks (accounting close); RAND Corporation (dispute adjudication). Authority enforcement source: traditional enforcement is application-specific - no universal proof mechanism exists; DID/VC verification: milliseconds; credential issuance: hours-days. Conduit benchmarks are internal - p50/p99 figures to be confirmed with engineering prior to publication.
The Smart Contract Problem
Smart contracts encode settlement logic. They do not encode legal agreements. Five structural problems make them unsuitable as the governing agreement between institutional counter-parties.
A smart contract encodes what its developer chose to encode. The governing agreement - representations and warranties, default conditions, dispute resolution mechanisms, regulatory compliance obligations - lives in a legal document the smart contract cannot read or enforce. The contract and the code are two separate instruments with no guaranteed relationship to each other.
Every deployed smart contract has a technical admin - typically the deploying organization - with the ability to upgrade, pause, modify, or terminate the contract. This authority is not disclosed in the legal agreement and was not granted by counter-parties. It exists as a technical capability outside the legal relationship. Counter-parties may not know it exists.
The smart contract executes what the code says, not what the parties intended. Bugs, mistranslations of legal language into code, and edge cases not covered by the developer do not constitute breach - they constitute the contract executing as written, regardless of what the parties agreed. The legal agreement and the executing code are different artifacts with no enforcement mechanism connecting them.
Real agreements change. Renegotiating a term, adding an obligation, or adjusting a covenant requires amending the smart contract. Amendment requires either redeployment (a new contract, losing state history) or an upgrade mechanism designed in at deployment time. Most smart contracts were not designed to be amended. Counter-parties who negotiate a change have no mechanism to reflect it in the executing contract without developer involvement.
Enterprise transactions operate across dependency graphs. One agreement's state change triggers obligations in downstream agreements. When a smart contract in this graph is amended or redeployed, every contract that depends on its state may break, produce incorrect outputs, or require manual intervention. There is no amendment mechanism that propagates correctly through dependent agreements by design.
How Conduit Addresses This
Conduit encodes the full legal agreement as a computable state machine - not settlement logic, but the complete agreement including obligations, conditions, governance rules, and milestone definitions. Amendments require counter-party sign-off, matching the legal amendment process. The state machine's transition rules are the agreement; there is no side document. Changes to an agreement propagate correctly through dependent agreements because the dependency structure is part of the specification.
The Authority Gap
Authentication proves who you are. Authorization grants access. Neither is authority. Authority is a verifiable claim that a specific action is within the scope of a specific delegation - traceable to the party who granted it, bounded by exactly what they permitted, and enforceable before the action executes. No traditional system and no blockchain implementation provides this.
TRADITIONAL SYSTEMS
Implied and Asserted, Never Proven
Traditional authority lives in two places: org structure and legal agreements. A job title implies certain signing limits. A board resolution asserts that a named officer may act within certain bounds. A power of attorney explicitly grants a defined scope.
None of these produce a proof that travels with the action. When an employee signs a contract, there is no mechanism embedded in the execution that verifies their signing authority against the scope granted to them. Enforcement is application-specific - a treasury system may have a hard limit; an ERP may require an approval workflow. But these checks are per-application, inconsistent across systems, and invisible to counter-parties.
When authority is violated - an employee exceeds their signing limit, an agent acts outside a mandate - it is typically discovered through reconciliation, audit, or dispute. The enforcement is reactive. There is no structural record of what scope was in force at the moment of execution, attached to the action itself.
BLOCKCHAIN
Authentication and Application-Coded Authorization
Blockchain introduced cryptographic authentication: a private key proves identity with mathematical certainty. Smart contract roles extended this to authorization: the owner role can call this function; others cannot.
This is a meaningful advance over traditional systems for the specific operations it covers. But it is not authority in the sense that matters for institutional transactions.
The authorization is binary. A smart contract role says you can call a function or you cannot. There is no mechanism to express: you may execute trades up to €1M in instruments of type X, between 09:00 and 17:00, with counter-parties in jurisdiction Y, until this delegation expires on date Z. This scope cannot be expressed at the execution layer. It can be coded into application logic, but that logic is specific to one contract, written by one developer, and changeable by whoever controls the contract.
There is no proof chain. Even when authorization works correctly, there is no verifiable record connecting the authorized action to the original human who granted the authorization. A wallet address has a role. Where that role came from, who granted it, under what conditions, and whether it was within the scope of what the grantor was themselves authorized to delegate - none of this is part of the execution record.
The authorization is not scope-expressive. There is no standard model for expressing what a party is permitted to do in terms that are machine-enforceable and legally meaningful simultaneously. DIDs and Verifiable Credentials can carry attributes and claims, but encoding institutional mandate scope as an enforceable, algebraically-constrained authorization model is not a solved problem in any deployed blockchain system.
CONDUIT: AUTHORITY AS A PROOF
Conduit's Authority operation is not authentication. It is not role-based access control. It is a cryptographic proof that a specific action is within the scope of a specific delegation, evaluated by Conduit OS before execution.
Every action on Conduit produces a provenance record tracing from the action through the delegation hierarchy to the original human grant of authority. This chain is not an audit log appended after execution - it is evaluated as a precondition. If the chain cannot be constructed and verified, the action does not execute.
The scope of what a delegate is permitted to do is expressed as a PowerScope - an algebraic capability model that defines the exact bounds of the delegation: instrument types, size limits, time windows, counterparty constraints, jurisdictional scope. PowerScopes compose correctly through delegation chains: a sub-delegate cannot hold a scope that exceeds what their delegator was permitted to grant.
Delegations are registered in Authority Domain cert registries - auditable, revocable, and verifiable by any party with the relevant public keys. Revoking a delegation propagates instantly through every sub-delegation it created. There is no session to expire, no cache to invalidate.
Conduit OS evaluates the authority proof chain before any state transition fires. An action outside the delegated scope is not flagged - it is rejected. There is no execution to reverse. The constraint is structural, not monitored.
The result: every action on Conduit carries an attached, verifiable proof of the authority under which it was taken. Adjudicating an authority dispute is a query against the event store, not a reconstruction of who had what title when.
The Capital Cost of Settlement Latency
Time-to-settlement is not just a latency metric. It is a capital cost. Every day between transaction execution and final settlement is a day that capital is encumbered - unavailable for other use, carrying an opportunity cost, and exposed to counterparty risk.
Illustrative Scenario
Mid-market trade finance company
Outstanding receivables
$50M
Net-90 terms
Cost of capital
7%
annually
Capital cost from latency
$863K / year
$50M × 7% × (90÷365)
This figure represents only the cost of capital encumbered by settlement latency. It does not include system integration and maintenance costs, reconciliation labor, dispute resolution, or counterparty risk exposure during the settlement window.
Under Conduit, settlement is part of the transaction. The capital is released at finality - under 100 milliseconds from execution. The $863K/year capital cost from settlement latency approaches zero.
| Outstanding receivables | Net-90 capital cost @ 7% |
|---|---|
| $10M | $172K / year |
| $50M | $863K / year |
| $250M | $4.3M / year |
| $1B | $17.3M / year |
Settlement latency costs only. For large institutional operators, the full cost of fragmented operations - system integration, reconciliation, disputes, and capital - is measured in tens to hundreds of millions annually.
The Adjudication Problem
The latency table shows what five operations cost when they run correctly. The adjudication problem is what happens when they don't - or when parties disagree about whether they did.
TRADITIONAL SYSTEMS
Reconstruction
In traditional finance, a disputed transaction means reconstruction. Each party holds their own records: the custodian's log, the ERP's accounting entries, the contract management system's version of the agreement, the settlement rail's confirmation. These systems were not designed to agree with each other. They agree when every integration fires correctly and every reconciliation runs clean. When they disagree, establishing ground truth requires forensic accounting, legal discovery, expert witnesses, and - in contested cases - litigation or arbitration.
The timeline: commercial arbitration under AAA rules takes six months to two years for cases that reach a hearing. Commercial litigation takes one to five years. Capital is locked for the duration. The parties remain exposed to each other - credit risk, market risk - until resolution. The cost is not only legal fees. It is the deals that never happen because sophisticated counterparties will not enter structures where the adjudication risk is too uncertain or too expensive.
BLOCKCHAIN
The Governing Agreement Problem
On blockchain, on-chain execution is deterministic and auditable. A dispute about on-chain state is often resolvable - the ledger is the record. But institutional transactions have governing agreements that live off-chain. When the smart contract and the legal agreement conflict - the code executed one outcome, the parties intended another - the dispute enters traditional litigation, supplemented by blockchain forensics.
Courts are interpreting smart contract code as legal instruments in real time, without established precedent in most jurisdictions. The evidentiary picture is worse than in traditional finance: there are now two authoritative records that can contradict each other, and no framework for resolving which governs.
CONDUIT: A DISPUTE IS A QUERY
On Conduit, the governing agreement is the executing state machine. There is no off-chain document that can conflict with the on-chain execution - they are the same artifact. The state machine's transition rules are the agreement terms. Every execution produces a provenance record as a precondition of execution, not an audit log appended afterward.
Adjudicating a dispute is a query: What was the agreement state at time T? What authority proof was evaluated? What transition fired and under what conditions? What rights were held and by whom? What was every counterparty's accounting position? The event store is authoritative and immutable. There is no reconstruction - the record is the record.
This does not eliminate all disputes. Interpretive questions - what did we intend when we agreed to this term? - still require human judgment. But the evidentiary basis for most financial disputes (what happened, in what order, authorized by whom, recorded where) is resolved by the architecture before litigation is ever considered. What remains is a narrower category of genuinely interpretive disagreement, with a complete, unambiguous factual record already established.
Private Alpha. Access by invitation.
We're opening access in stages. Tell us what you're evaluating and we'll reach out when it's a fit.