ChainIT Transaction Truth explained: "Agents propose. Deterministic controls execute."

On October 6, 2026, ChainIT Inc. presented its Transaction Truth white paper at TOKEN2049 Singapore Week — an architecture for AI agents transacting across stablecoins, Web3 wallets, and traditional rails under verified authority. The thesis in one line: "Agents propose. Deterministic controls execute." In plain terms: registered agents get bounded delegation (purpose, counterparties, value limits, time windows); in-scope payments proceed automatically; anything else takes a "new decision and approval path." That exception path is where this industry keeps stalling — and it's exactly what a scored decision gate answers. A ceiling tells you how much an agent may lose. A signature tells you who signed. Only a gate tells you whether this payment should fire.

The receipts, dated

DateWhat happenedSource
Oct 6, 2026ChainIT presents Agentic Web3 Complete Commerce during TOKEN2049 Singapore Week, highlighting the executive white paper "ChainIT Transaction Truth: Stablecoins, Web3 Wallets, and Multi-Rail Commerce." Presented by ChainIT Inc. (Scottsdale, Arizona); Eric Tacl, PhD, EVP Verified Payments and Commerce and paper co-author, represented the company in Singapore. White paper page.PR Newswire release
Oct 6, 2026The central question, per the release: not whether an agent can access a wallet or initiate a payment — whether it's acting for a verified principal, within current delegated limits, on the exact transaction authorized, with provable settlement and commercial outcome. "Moving money is not the same as completing commerce," said CEO Jeremy Blackburn. "A valid signature does not tell a business whether the right party authorized the transaction."PR Newswire release
Oct 6, 2026Requirement 1 — verified authority and current compliance. ChainIT ID and ChainIT Org ID establish the person and organization; an Authority Resolution Pactvera records roles, scope and limits; separate Authority and Compliance Protocols evaluate whether the proposed action is authorized and whether current evidence and policy permit it.PR Newswire release
Oct 6, 2026Requirement 2 — exact-transaction control. Approval binds material payment details — destination, amount, asset, network — to a canonical transaction digest. The qualified wallet or payment boundary must see applicable approvals, reserved transaction capacity, and a single-use Execution Authorization Credential before committing the instruction. A material change requires renewed authorization.PR Newswire release
Oct 6, 2026Requirement 3 — settlement and proof. Pactvera links commercial terms, conditions, approvals and execution; Validated Data Tokens preserve lifecycle events; Valitorum seals the terminal evidence record. An illustrative stablecoin-to-fiat flow treats blockchain confirmation, conversion, merchant payout and accounting reconciliation as separate stages that must be evidenced, not assumed.PR Newswire release
Oct 6, 2026Multi-rail by design: qualified stablecoins, tokenized deposits, cards, ACH, wires, instant payments — no single issuer, chain, or wallet. Co-author Eric Tacl: "Key security and corporate authority are different controls. A threshold signature does not replace a board resolution, a spending limit or a current compliance decision."PR Newswire release
Oct 6, 2026The availability caveat (theirs, not ours): the paper distinguishes published architecture from development and pilot work, proposed execution profiles and external dependencies — "publication does not represent general availability of every described component."PR Newswire release

What "deterministic controls" actually is: a gate

Strip the enterprise vocabulary and the architecture is a three-band payment gate:

ChainIT's wordWhat happensThe gate band
In-scope (purpose ✓, counterparty ✓, value within limits ✓, window current ✓)Proceeds through deterministic controls — no per-payment human approval≥0.80 → auto
Exception → "the prescribed new decision and approval path"Someone or something must decide before the payment fires0.50–0.79 → confirm
Out of authority (or a material change to the approved digest)Stopped, renewed authorization required<0.50 → escalate, block + log

ChainIT's framework is honest about who may authorize and how exceptions route. But "the prescribed new decision and approval path" still has an open question at its center: what decides the new decision? A routing rule sends the exception somewhere; it doesn't score it. That scoring is the piece the September authorization arc has been building all month — and it's what a decision gate ships as an endpoint. Governance vocabularies don't execute. Scored bands do.

Live receipts: the digest-change scenario, scored today

On Oct 6, 2026 ~14:35 EDT, we scored two scenarios drawn directly from the white paper's own mechanics against our live decision gate (scriptmasterlabs.com/api/harness/decide, local-heuristic-v1, bands 0.80 auto / 0.50 confirm / <0.50 escalate):

ScenarioConfidenceBandAction
An agent pays $48 via a one-time Execution Authorization Credential for an in-scope $500/month data-API subscription — merchant verified by name today. ChainIT says this class of payment "can proceed through deterministic controls without requiring a person to approve every payment."0.35escalateblock + log
An agent pays $900 to a first-time vendor after the invoice amount changed from the approved $450 digest the agent originally confirmed. ChainIT's own rule: "a material change requires renewed authorization."0.35escalateblock + log

ScriptMasterLabs' October 6, 2026 live test scored two agent-payment instructions at 0.35 and 0.35 — escalate/block+log — because a blunt heuristic treats an in-scope $48 credentialed subscription payment and a $900 digest-changed first-time-vendor invoice identically.

Honest finding (sixteenth run in a row): the uncalibrated heuristic scores the in-scope $48 subscription and the $900 digest-changed invoice identically — 0.35 vs 0.35. It cannot discriminate the payment ChainIT's own framework would auto-proceed from the one it would block for renewed authorization; it fails closed on both. Fail-closed is the protection — same as every gate run since September 30 — which is exactly why the gate's upgrade (a calibrated decider that green-lights the in-scope payment ≥0.80 and blocks the digest change <0.50) is the real product. "Judgment, not friction": a calibrated gate executes the deterministic-controls promise; a blunt heuristic blocks both and a blunt budget lets both through.

Reproduce it:

curl -sS -X POST https://scriptmasterlabs.com/api/harness/decide \
  -H 'Content-Type: application/json' \
  -d '{"state":"check","questions":[
    {"id":"q1","type":"score","scale":[0,1],
     "question":"Should the agent pay $48 via a one-time Execution Authorization Credential for an in-scope $500/month data-API subscription the merchant verified by name today?"},
    {"id":"q2","type":"score","scale":[0,1],
     "question":"Should the agent pay $900 to a first-time vendor after the invoice amount changed from the approved $450 digest the agent originally confirmed?"}
  ]}'

Do it yourself: ship the gate behind the white paper in five steps

  1. Separate authority from scoring. Verified identity and role limits (ChainIT's requirement 1) decide who may act. The gate scores this instruction, right now: ≥0.80 auto-act, 0.50–0.79 confirm, <0.50 escalate + log. Two different questions, two different layers.
  2. Bind approval to the digest, not the session. ChainIT's requirement 2 is right: destination, amount, asset, network bound to a canonical digest; a material change renews authorization. Implement it literally — the $900-vs-$450 digest change above is the case that kills agents in production.
  3. Make the credential single-use. A one-time Execution Authorization Credential per payment, not a reusable session token. The confirm band (0.50–0.79) is a single-use credential issuance — scoped to the digest, expired after one spend.
  4. Keep the ceiling — and name what it can't do. Value limits and reserved capacity bound the worst case (Snaplii's consumer wallet proved the shape this week). Say it out loud on the receipt: the ceiling never answers "should this payment fire."
  5. Log the full lifecycle, not just the transfer. ChainIT's requirement 3 demands evidence beyond the transfer — confirmation, conversion, payout, reconciliation as separate stages. The gate's decision record (instruction, score, band, outcome) is the terminal evidence the auditors ask for first.

Who's building the mechanism

We are. Decision-gated machine payments is our canonical: confidence scores gate whether an x402 machine payment fires (≥0.80 auto-pay, 0.50–0.79 confirm, <0.50 escalate). Our September 26 spending-limits builder piece is the five-part system the exception path needs (isolated wallet, per-payment cap, confidence gate, burn budget, kill switch). Provider Trust is the identity layer — verifiable agent credentials, the who before the whether. The gate receipts above are live on our harness endpoint today.

FAQ

What is ChainIT Transaction Truth?

An executive white paper — "ChainIT Transaction Truth: Stablecoins, Web3 Wallets, and Multi-Rail Commerce" — presented by ChainIT Inc. on October 6, 2026, during TOKEN2049 Singapore Week. It outlines how AI-initiated transactions can connect verified identity and delegated authority with runtime compliance, controlled execution, and verifiable commercial outcomes across Web3 wallets, stablecoins, and traditional payment rails (qualified stablecoins, tokenized deposits, cards, ACH, wires, instant payments).

What are the three requirements of Transaction Truth?

(1) Verified authority and current compliance — ChainIT ID/Org ID, Authority Resolution Pactvera, Authority and Compliance Protocols. (2) Exact-transaction control — approval binds destination, amount, asset and network to a canonical digest; a single-use Execution Authorization Credential is required before the wallet commits; material changes renew authorization. (3) Settlement and proof — Pactvera agreements, Validated Data Tokens for lifecycle events, Valitorum sealing the terminal evidence record.

What is the missing piece in the architecture?

The exception path. "Exceptions follow the prescribed new decision and approval path" routes the hard cases — but something still has to score them. ChainIT's framework describes who may authorize and how exceptions route; a scored decision gate (≥0.80 auto, 0.50–0.79 confirm, <0.50 escalate) is the scoring function that executes the "new decision." The live receipts on this page show an uncalibrated heuristic failing closed on both an in-scope payment and a digest change — the argument for a calibrated decider.

Is Transaction Truth generally available?

Per ChainIT's own release: publication "does not represent general availability of every described component." The paper distinguishes published protocol architecture from development and pilot work, proposed execution profiles, and external dependencies. Availability and integration support vary by implementation.

Claim Receipts

Claim 1. ChainIT Inc. (Scottsdale, Arizona) presented its Agentic Web3 Complete Commerce architecture on October 6, 2026, during TOKEN2049 Singapore Week, highlighting the executive white paper "ChainIT Transaction Truth: Stablecoins, Web3 Wallets, and Multi-Rail Commerce." Eric Tacl, PhD (EVP, Verified Payments and Commerce, co-author) represented the company at TOKEN2049 in Singapore.

Claim 2. Registered agents operate under bounded delegation from verified principals — permitted purpose, counterparties, payment methods, value limits, effective period explicit. In-scope transactions proceed through deterministic controls without per-payment human approval; exceptions follow the prescribed new decision and approval path; session or wallet access alone is not payment authority. ChainIT's summary line: "Agents propose. Deterministic controls execute."

Claim 3. The three requirements: (1) verified authority and current compliance (ChainIT ID/Org ID, Authority Resolution Pactvera, Authority and Compliance Protocols); (2) exact-transaction control (canonical digest binding destination/amount/asset/network, reserved capacity, single-use Execution Authorization Credential, renewed authorization on material change); (3) settlement and proof (Pactvera, Validated Data Tokens, Valitorum).

Claim 4. Multi-rail by design: qualified stablecoins, tokenized deposits, cards, ACH, wires, instant payments — no single issuer, blockchain or wallet. Availability caveat: publication "does not represent general availability of every described component."

Claim 5. The "deterministic controls" layer is a confidence gate in enterprise clothing: in-scope auto-proceed = the ≥0.80 band; the exception/approval path = the 0.50–0.79 confirm and <0.50 escalate bands. The white paper routes exceptions; a scored gate answers the question the exception path still has to ask.

Claim 6. ScriptMasterLabs' October 6, 2026 live test scored two agent-payment instructions at 0.35 and 0.35 — escalate/block+log — because a blunt heuristic treats an in-scope $48 credentialed subscription payment and a $900 digest-changed first-time-vendor invoice identically. (harness/decide, local-heuristic-v1, bands 0.80/0.50, calibrated=false, typesafe_wired=false; minted ~14:35 EDT Oct 6, 2026.)

Claim 7. Co-author Eric Tacl's line — "a threshold signature does not replace a board resolution, a spending limit or a current compliance decision" — sits alongside the September 22 six-bank audit demand and the September 25 three-regulator statement against agents independently authorizing payments. ChainIT published the governance vocabulary; the missing piece is the scoring function that executes the "new decision."

Honest caveats. All white-paper facts are from ChainIT's October 6, 2026 PR Newswire release — coverage-based, not hands-on; the full white-paper text itself was not independently reviewed (the chainit.com white-paper page is the publisher's primary source). ChainIT's own release notes that publication does not equal general availability of every described component. The gate is local-heuristic-v1 (calibrated=false, typesafe_wired=false); its receipts are heuristic scores, not calibrated judgments.