A parametric gap quote · the cost of an unhedged liability
Someone made a claim in public that does not survive being checked. This hands you a way to say so that is neither a shrug nor a shout: a correction with a receipt attached, and a suggested remedy that is a real product purchase, not a payment to a person. Nothing here creates a debt. It is an appeal to look.
A public claim with no denominator. The oldest charge, and the easiest to check.
Material connection is disclosed, not hidden — the honest path is the default; true independence is the one you assert.
Both carry the same pay instruction: the payment is made at thetadriven.com/pricing, and the payment generates the license. An invoice corrects; a quote prices — your call.
ADVISORY QUOTE — a price for a liability you hold
This is an advisory quote. It does not allege a wrong you committed — it prices a liability you already hold: your agent-year count against the generative-AI exclusion attaching at your renewal. An offer to be counted, not a demand.
Generated by an Independent Verifier, under their own name. The ledger does not require a pre-existing relationship to recompute a public claim.
This document creates no legal debt, because you have not yet secured the license to offload your liability. No contract exists, and nothing has been rendered. This is a quantified read of your public Trust Debt — a checkable, public claim, not an invoice in the legal sense. Paying the exact parametric gap is how you cryptographically close it; ignoring it leaves the gap on the public ledger.
To: [recipient]
The public statement being corrected (quote it exactly, with a link or screenshot as the receipt):
[paste the exact public sentence here]
The correction:
You stated, in public, that your agents are verified / safe / under control. That is a claim with a denominator, and it does not have one yet: run the recompute below on the exact system you described, and it returns a signed placement verdict a stranger can re-run. If it lands in-lane, publish that and we will amplify it. If it does not, the claim needs a receipt it currently lacks. (If the specific slip was the word "certain" or "deterministic", the why is a footnote, not the charge: thetadriven.com/blog/2026-08-13-the-river-is-the-prompt.)
The receipt — verify it yourself:
In your browser, nothing to install: thetadriven.com/verify-receipt (signature check plus a full recompute). Or, if you can run packages on this machine: npx -y thetacog-mcp@latest attest-demo (runs locally, nothing leaves your machine).
The price, derived from your agent-year count — not a number we chose:
You run 3 agent-years. At $20 each that is $60. Payable only at thetadriven.com/pricing — the site's ordinary credit-card checkout. There is no other payment channel, no wire, no invoice number: the page is the whole transaction, which is also what makes it checkable. A license is an attestation key, delivered by email, metered at 365 days or 10,000 attestations, whichever comes first.
Reading, before you decide anything:
Tesseract Physics — Fire Together, Ground Together (thetadriven.com/book) · the adoption board, unflattering numbers included (thetadriven.com/benchmark) · why the word misleads (thetadriven.com/blog/2026-08-13-the-river-is-the-prompt)
Why you are hearing this from an unusual channel:
This quote was not generated to insult your engineering. It was generated because the math did not resolve. The standard is indifferent to intent — it only reads the state-proof. The target is the claim, never you: a market that keeps selling confidence nobody can back with a count.
Disclosure: I use ThetaDriven's instrument; I am not employed by them. (If you have any closer relationship — contributor, license holder, program operator — state it here instead.)
From: [your name or handle]
Subject
A public claim you made that does not have a denominator yet
Body
Hi [name], You said in public that your agents are verified and under control. That is a claim with a denominator, and I do not think it has one yet. You do not have to trust me on that. Open the attachment and run the one line inside — in your browser, or on your own machine with nothing leaving it. It returns a signed placement verdict a stranger can re-run. If it lands in-lane, publish that and I will amplify it. If it does not, the claim needs a receipt it currently lacks, and the attached shows what producing one costs from your own agent count. Nothing is owed to me. There is no debt, no relationship, and no obligation — this is an appeal to look. — [your name] Disclosure: I use ThetaDriven's instrument; I am not employed by them. (If you have any closer relationship — contributor, license holder, program operator — state it here instead.)
Attach the .docx above. Send it under your own name, from your own address, never on any company's letterhead — and one recipient at a time, because a mass-send is the thing that turns a correction into a campaign you did not mean to run.
The only payment control on this page. It opens the real thetadriven.com/pricing checkout — no invoice number, no wire, no debt.
Before you send anything, check the object it references — in the browser, nothing installed: /verify-receipt, or on your own silicon: npx -y thetacog-mcp@latest attest-demo
The mechanics under the document — Trust Debt (the recomputable gap between what an agent was specified to do and what it actually did), what a breach pays out, warranty versus insurance, and why only the attestation ever leaves the building — in full:
The quantifiable delta between what an agent was specified to do and what it actually, ungroundedly did — and it is permanently on the record. Every ungrounded action adds to the balance. That is the object being priced here: not a vibe, a difference you can recompute.
An opaque, unverified model is an uninsurable risk — the market has already carved these losses out of coverage. Run agents on one and the enterprise holds a ruinous, unpriced liability, because nobody can count which step failed. That is the whole reason a recomputable count is worth paying for.
A failed state-proof is the trigger: the sealed reading crosses the band written into the paper, no adjuster's judgment required. The payout is remediation capacity — GPU-hours and engineering hours, dispatched through an accredited network to whoever holds the highest-confidence pixel over the failure region, covering the lost compute and consulting hours. It is a repair, not an open-ended cash indemnity, which is what keeps the paper in warranty territory. The cash rails, options, bonds and derivatives that let an organization hedge foundation-model volatility are real, and they are a sponsor's to structure downstream of the print, inside their own regulatory perimeter — never operated from here. We license the measurement; we do not run the pool.
Selling the SLA as a warranty versus insurance comes down to one thing: how you get paid. Buy the warranty from us and the payout is in kind — our supplier network makes you good, paying the consultants and the compute directly, so the failure gets fixed rather than merely reimbursed. Want cash instead? Go through a broker. A large insurance broker can wrap insurance on top of the same measurable reading and pay out in cash, carrying the underlying liability itself — which is exactly why it wants to: the telemetry lets it open substantial new premium pools without holding a risk it could not otherwise price. Either way the reading is identical. We license the measurement, the broker carries the liability, and you pick the payout.
No proprietary data leaves the building. Only the attestations — provably decidable, and recomputable by anyone who doesn't trust them — are pushed to the immutable public GitHub ledger, where they stand as the proof a claim references. The receipt is the whole transaction, which is also what makes it checkable. npx -y thetacog-mcp@latest attest-demo
Decidable is a higher standard than interpretable. Interpretability is a qualitative read; decidability turns an output into an objective, discrete, recomputable state — the only kind of thing an automated warranty can trigger on. Tagging output with a model adds metadata; it does not make the record decidable.
Can't run a package on a work machine? Verify a signed receipt in the browser — signature check plus a full recompute, nothing installed: /verify-receipt.
This page composes a correction for someone else's public claim. To invoke your own coverage, the failing commit's receipt is the claim reference — there is no separate claims form, and that is the point. Run npx -y thetacog-mcp@latest attest-demo on the failing commit, verify the signed receipt at /verify-receipt, and the coordinate it returns is what dispatch reads to route in-kind restoration — consultant and GPU hours — to whoever holds the highest-confidence pixel over your failure region. Want cash instead of restoration? That path is your broker's, on the same reading.
One honest caveat, because this is the single paragraph on the page where we would take on a duty rather than disclaim one: this describes the intended mechanism, and the dispatch network's capacity is committed per pool in the T5 hardening documents, not implied here. Whether a paid promise of remediation on a triggering event is a regulated service contract is a state-by-state, counsel-reviewed determination — being in kind does not automatically exempt it. No coverage is bound by this page; the license is a measurement license, and any pool that pays on it is structured by its sponsor under its own counsel.
This is one rung of a larger, disclosed process. The full escalation ladder — capped so it never names a private party — is the playbook. The argument for why the correction matters at all is the pamphlet. The single link you send the recipient alongside this invoice — so they can understand every implication for themselves — is the resource pack.
Run the receipt yourself before you send anything: npx -y thetacog-mcp@latest attest-demo
