The decision endpoint
din.org renders reasoned, auditable rulings on the disputes, claims, and decisions you no longer want to make in-house — from a contract dispute between two people to a marketplace refund to a public procurement award. File the case. Receive the verdict. Ship the outcome.
Used internally by din.org's own AI arbitration tribunal · In design with public-sector pilots, marketplaces, insurers & institutional tribunals
Pilot partners and reference implementations
Three principles. In this order.
I
The law
Statute, contract, regulation, platform policy — whichever rule set governs. The first question is always whether the ruling is defensible on legal grounds. If a higher court reviewed it, would it stand?
II
The parties
Within what the law allows, what serves the actual humans involved — the buyer and the seller, the claimant and the insurer, the citizen and the agency? Fairness is not the same as formal correctness.
III
The whole
Decisions create precedent. A ruling that punishes one bad actor changes incentives for thousands. The third principle keeps every individual ruling tethered to the system it helps shape.
The order matters. Law first protects legal defensibility. Parties second protects fairness. Society third anchors every ruling to the world it lives in.
For your own dispute
A contract dispute. A defective product. A landlord refusing to return a deposit. An inheritance disagreement. A B2B contract that's going sideways. File it directly with din.org — same engine, same three principles, same Big Four audit, same appeal path. No API integration required.
You invite the other party. Both submit evidence. The engine renders a reasoned ruling, days — not years. Either side can appeal to a human-judge panel. Enforceable internationally under the New York Convention 1958, recognized in 170+ jurisdictions.
Civil disputes — direct route
Both parties must consent to arbitration · or a clause must exist
Who uses it
Public sector & high-integrity decisions
In development with municipal pilots
The decisions where conflict of interest matters more than speed. Every input recorded, every step auditable, every ruling appealable. Designed for jurisdictions where the perception of fairness is as critical as fairness itself — and where the absence of an independent third party historically distorts public outcomes.
Typical use: 10 — 5,000 decisions / month
Insurance & banks
Pilots in DACH
Adjusters use din.org for first-pass decisions on contested claims. The reasoned ruling becomes the artifact the customer sees — not the adjuster's name. Speeds up cycle time, removes the personal-blame surface, gives the regulator a paper trail.
Typical use: 1,000 — 250,000 decisions / month
Marketplaces & platforms
Pilots in design
Trust & safety teams escalate edge cases to din.org instead of arguing with two upset users. Every refund denial, every seller suspension comes with a reasoned, third-party ruling the user can read and challenge — not a customer-service form letter.
Typical use: 200 — 50,000 decisions / month
Institutions & tribunals
Reference partners
Existing institutions integrate din.org as their AI tier — for triage, first-instance rulings, parallel sanity-checks against human decisions. The institution retains brand and authority; din.org provides the engine, speed, and audit trail.
Typical use: 50 — 5,000 decisions / month
How a decision looks
curl https://api.din.org/v1/decisions \
-H "Authorization: Bearer sk_live_..." \
-H "Content-Type: application/json" \
-d '{
"framework": "marketplace_dispute",
"jurisdiction": "DE",
"claim": {
"type": "refund_request",
"amount": { "value": 1240, "currency": "EUR" }
},
"parties": [
{ "role": "buyer", "id": "u_2k9b8m" },
{ "role": "seller", "id": "m_71xx3z" }
],
"submissions": [
{ "from": "buyer", "text": "Headphones arrived with..." },
{ "from": "seller", "text": "Item matches the listing..." }
],
"evidence": [
{ "type": "image", "uri": "..." },
{ "type": "chat_transcript", "uri": "..." },
{ "type": "listing_snapshot","uri": "..." }
]
}'{
"decision_id": "dec_8h2k4mqzWp0Lt9Rx",
"ruling": "partial_refund",
"amount_awarded": {
"value": 620, "currency": "EUR"
},
"principles_applied": {
"law": "marketplace_policy_4.2.1, BGB §437",
"parties": "buyer harm vs. seller good-faith listing",
"society": "incentive against misrepresentation,
discouraging stale claims"
},
"reasoning": "The listing describes the item as
'as new — used twice'. Buyer evidence shows
visible cosmetic wear inconsistent with that
representation. However, the buyer waited 22
days before raising a claim, exceeding the
platform's 14-day clear-defect window.
A 50 % refund balances the misrepresentation
against the buyer's delayed escalation.",
"citations": [
{ "source": "marketplace_policy_4.2.1", "weight": 0.81 },
{ "source": "BGB §437", "weight": 0.42 }
],
"confidence": 0.83,
"appeal_eligible": true,
"audit_trail_url":
"https://din.org/audit/dec_8h2k4mqzWp0Lt9Rx",
"issued_at": "2026-05-13T14:22:09Z"
}Reasoned
Every ruling ships with the chain of reasoning, weighted citations, the three principles weighed against each other, and a confidence score — not a black-box verdict.
Auditable
Each decision has a signed audit trail URL. Inputs, intermediate steps, and final ruling are reproducible on replay. See "Open to audit" below.
Appealable
Every ruling is appeal-eligible. Appeals go through a separate human-judge panel — see Pricing.
The economics of care
No queue priority by case value. No "small cases get shorter reasoning." One decision pipeline, identical rigor, identical depth — regardless of who is filing or what is at stake. The marginal cost of justice approaches zero. We allocate the same care to everyone.
Why outsource the verdict
An internal "AI decision tool" is still your decision. An adjuster, a moderator, a committee, a procurement officer. The customer or the citizen still hears it from you. din.org is the structural opposite: a third party seen as the source of the ruling — with the audit trail, independence, and appeal mechanism to back it up.
01
Your decision isn't your decision anymore. The ruling comes from din.org with a verifiable audit URL the affected party can inspect. You executed; we decided.
02
Every input, every intermediate step, every citation is recorded and signed. If a regulator, court, or appeal panel asks for the basis, you hand them a URL — not a deposition.
03
Every ruling is appeal-eligible. Customers and citizens don't escalate to your CEO; they escalate to a din.org human-judge panel of 1, 3, 5, or 7 jurors — sized for the stakes. The conflict leaves your organization. See how the appeal works.
A quiet artifact
"Per the attached ruling,
this is the outcome."
The appeal
Every decision rendered by din.org — base ruling, verified proceeding, civil-dispute verdict, public-sector decision — is appeal-eligible by default. Appeals leave the AI tier entirely. They are decided by a panel of human judges drawn from a vetted roster of trained arbitrators. The panel reviews the record, considers new submissions, and issues a final, binding ruling.
Solo judge
1 juror
Fast-track. Low-value civil disputes, administrative offences, simple consumer claims. Days from filing to ruling.
Standard chamber
3 jurors
The default appeal panel. Majority decides. Suitable for most commercial, marketplace, and institutional disputes.
Expanded panel
5 jurors
For high-value, technically complex, or cross-border disputes. Specialist judges included where the matter requires.
Full bench
7 jurors
Matters of public significance, precedent-setting cases, or sensitive public-sector decisions. Maximum legitimacy.
Who decides
Vetted human arbitrators — trained, conflict-screened, and rotated. Each panel includes specialists where the matter requires it (commercial, public-procurement, technical, etc.).
What they review
The full din.org record — inputs, evidence, the three-principle weighing, the AI ruling, and the audit trail — plus any new submissions either party files on appeal.
How it ends
A final, binding ruling with full reasoning. Enforceable internationally under the New York Convention 1958 in 170+ jurisdictions, or recognized by the institution operating the ruling.
Panel size is chosen by the appealing party — or by the platform when stakes or complexity require it. Appeal pricing scales with panel size and case value. The right to appeal is not optional; it is the architecture.
Open to audit
The decision mechanism — model architecture, training corpus, weighting model, per-case reasoning trails, the application of the three principles — is subject to independent annual audit by a Big Four accounting firm. Clients receive the full audit report on request, including findings, exceptions, and management responses.
What is audited
Architecture · training data sources · weighting model · per-case audit trails · principle application · appeal frequency · disagreement rate with human reviewers.
Who audits
A Big Four firm under formal audit engagement. Auditor identity disclosed in the report. Annual cadence. Findings published to clients verbatim.
Why this matters
The question isn't "can the AI decide?" — it's "can someone verify the AI decided correctly?" The audit is what makes the ruling defensible in front of a regulator, a court, a board, or a public.
We sell a mechanism, not a marketing claim. Anything we say about how a din.org decision is reached should be verifiable by a third-party auditor. That is the entire business model.
Where it's used
Civil contract disputes
private · B2C & B2B
Tenancy & landlord disputes
private · housing
Defective-goods consumer claims
private · retail
Small-claims arbitration
private · low value
Public procurement / tender awards
public sector · high-integrity
Grant & subsidy allocation
public sector · selection
Building permit appeals
public sector · regulatory
Public-office candidate vetting
public sector · integrity
Whistleblower complaint review
public sector · protected
Electoral & campaign-finance disputes
public sector · sensitive
Insurance claim adjudication
insurance · first-pass
Insurance fraud-flag review
insurance · automated triage
Bank complaint resolution
banking · consumer protection
Marketplace refund disputes
platform · consumer
Marketplace trust & safety appeals
platform · seller bans
B2B contract disputes
commercial · cross-border
Sports & federation tribunals
institutional · eligibility
Consumer ombudsman cases
institutional · regulated
HR disciplinary first-instance
internal · neutralization
Academic plagiarism / exam appeals
institutional · academic
Scholarship & award juries
institutional · selection
IP & patent first-look
commercial · technical
Inheritance pre-assessments
private · non-binding
Regulatory & compliance escalations
regulatory · audit
Same API. Different framework parameter. If your decision domain isn't listed, we've probably already designed a framework for it — ask.
The sandbox
The sandbox runs your real case end-to-end on the same engine that produces production rulings. Submit the facts, attach evidence, present your strongest argument and your opponent's — the engine returns a full reasoned ruling, the principles it weighed, and the confidence behind every step.
No commitment. No integration required. The same audit trail you would receive in production — but the outcome is yours to keep, act on, or discard.
Sandbox · what you get
Pricing
Issuing a reasoned decision should be commodity. The hard parts — human appeal, identity verification, video proceedings — are where we charge. Most integrations stay free.
Base decisions
Free
Up to ~1,000 tokens per decision
Overage billed by token thereafter
Appeals · where we earn
Per appeal
Human-judge panel · second instance
Tiered by case value · custom for institutions
Verified proceedings
Per case
Identity check · video hearing
Use when stakes demand it
Reference implementation
din.org operates the world's first end-to-end AI arbitration tribunal — case filing, examination, ruling, and appeal — entirely on top of this decision endpoint. You can see how the engine performs at full depth before you wire it into your own product.
What comes next
The engine that decides a marketplace refund is structurally the same as the one that could decide a parking dispute, a small-claims matter, or — eventually, where law permits — a lower-instance petty-offence case. Three principles, audited mechanism, mandatory human-judge appeal. We extend the surface only where regulators invite it.
01
In regulatory dialogue
Many jurisdictions are piloting AI-assisted first-instance review for high-volume, low-value cases. din.org is designed for it — engine, audit trail, mandatory appeal layer, all aligned with what such pilots require.
02
Architecturally ready
Speed-camera disputes, parking appeals, low-level municipal penalties. High volume, narrow factual scope, clear statute, low harm of error — the natural testing ground for AI adjudication with strong appeal rights.
03
Where law permits
Some jurisdictions are opening regulated experimentation for AI-assisted first-instance decisions in petty criminal matters — under strict appeal guarantees and human oversight. Where they do, the same engine — same three principles, same Big Four audit, same mandatory human-judge appeal — is built to participate.
din.org makes no claim to replace criminal courts. Where the architecture is invited into regulated experimentation, it is built to participate transparently — and to be removed transparently if a jurisdiction decides it no longer should.
Request access
Early-access partners get full integration support, a dedicated decision framework for their use case, and lifetime volume terms on appeal and verification billing.
Already have an account? Sign in here.