Decentralised News · Agentic Finance · Decision research
By Heath Muchena · Source review 2026-11-11 · Instrument version 1.0.0
What Matters
An agent needs to know what research it may use, for whom, in which outputs and for how long. Those permissions must remain distinct from update promises, prices and remedies. DN’s nine-part canvas turns a proposed licence into a reviewable record, while preserving unknowns and the authoritative text. It is a drafting aid, not a newly adopted DN contract.
A buyer can understand a report and still have no reliable answer to a basic question: may my team’s agent place this material in a customer-facing answer? “AI-ready” does not resolve that question. Nor does a downloadable JSON file that contains a price but omits redistribution conditions.
The practical product is a consistent pair: readable terms that state the bargain and a structured record that carries the same meaning into a workflow. Build both from the same reviewed version. This canvas helps a seller specify that bargain and helps a buyer identify the questions still blocking purchase.
DN Evidence Block
Sources reviewed: 2026-11-11 · Evidence type: W3C specifications and primary public-licence text, used as bounded reference points for an original procurement canvas.
Scope: Nine business-term groups for one proposed research offer. This article does not set DN’s product prices, entitlement rules or service guarantees.
- Policy expression: W3C ODRL 2.2 models permissions, prohibitions, duties and constraints. Encoding a policy does not establish that a supplier owns the rights it purports to grant.
- Version identity: W3C DCAT 3 provides vocabulary for dataset metadata and version relationships. A stable asset identifier and a specific version answer different questions.
- Scope matters: CC BY 4.0 illustrates a defined grant with conditions and limits. A custom research offer needs its own reviewed terms rather than an invented badge.
Not established by this work: No claim that a named shopping agent can interpret this canvas, that the export conforms to ODRL or DCAT, or that a completed canvas forms an enforceable agreement.
Original DN instrument · version 1.0.0
DN Research Licence Canvas
Prepare a nine-part term record for one proposed offer. The output identifies incomplete and contradictory groups; it does not create or approve a contract.
Enter your assumptions or load a labelled example. Required blank numbers stay missing; unknown evidence remains unknown. Use nonpersonal labels only. This instrument makes no network requests and uses no persistent browser storage.
Use your own inputs
No current result. Complete inputs or load an illustrative example.
The canvas checks the presence and stated status of nine term groups. It does not verify rights ownership, contract authority, legal adequacy, price fairness, technical interpretation or performance. It is not an adopted DN licence.
1. Which assets and versions are covered?
Identify the exact report, dataset, record range, attachment or service. State the version or effective date and flag excluded material. “Our research” is too broad when the package contains third-party charts, confidential interviews or components with different terms. A buyer should be able to match the delivered file to the item named in the offer.
Make it inspectable: give the asset a stable identifier and record the specific edition supplied. State whether attachments, source snapshots and later revisions are included. A product name alone may be too broad to tell a buyer what its entitlement covers.
2. Which buyer, people and agents may use them?
State which legal entity buys the entitlement and which people or agents may use it. Explain whether contractors, affiliates, service accounts and hosted processors fit within that permission. Do not use a seat count as an accidental substitute for a delegation policy.
Make it inspectable: list permitted roles and any relevant recipient limits. Identify the responsible organisation without exposing credentials or personal account details. An undefined “team” should not quietly become unlimited downstream distribution.
3. Which activities are permitted or prohibited?
Separate reading, internal analysis, retrieval, quotation, transformation, model training and delivery to customers. Define the intended use in ordinary language before mapping it to structured values. A permission to use a research answer internally does not itself describe the rights needed to resell an answer service.
For example, a hypothetical offer could permit a named company to retrieve specified records for an internal procurement assistant while excluding redistribution of the underlying dataset. That is a possible commercial design, not DN’s current offer. The contract must state its actual boundaries, and the workflow must preserve them.
The useful buying question is whether the licensed scope completes the buyer’s job. A broad-sounding label that leaves the intended output unresolved is a procurement gap even if the price is attractive.
4. Which outputs may leave the organisation?
Distinguish a short cited finding, a client deliverable, a downloadable table and a substitute research database. State what customers or partners may receive, what notices must accompany it and whether further reuse is allowed. Quantitative limits should identify their unit and measurement window; a limit called “reasonable use” needs enough context to guide a decision.
These details also affect price. A permanent archive entitlement, an internal retrieval right and permission to power a resale product are different bundles. Charging them differently can be coherent if the difference is visible before purchase and accurately represented after it.
5. What happens to retention, caches and copies after termination?
Explain what happens to downloaded files, caches, embeddings, backups and previously delivered outputs when access ends. The promise to stop future API access is different from a promise that every downstream copy has disappeared. If the intended arrangement requires deletion, specify its scope, process and any retained evidence or permitted exceptions.
A fictional licence might allow an internal cache during an active subscription and require deletion of that cache after termination, while allowing a previously issued client report to remain in the client’s records. Another licence might take a different approach. The canvas records the terms offered; it does not infer either design from the word subscription.
Make the technical obligations testable. Which system contains the licensed material? Who controls its retention? What does an offboarding record show? A buyer should establish this before promising its own customer that all supplier data can be removed on demand.
6. What coverage, updates and delivery are promised?
State what is included now, what will be maintained, the review cadence if any, and the conditions that can end coverage. A right to use a snapshot is not a promise that it stays current. “Updated regularly” leaves an agent unable to decide whether an old answer can still support a purchase.
Make it inspectable: distinguish a snapshot from a maintained service. Identify the fields or claims maintained, the notification route and any explicit exclusions. A possibility of future updates is different from a documented delivery obligation.
7. How do corrections and historical records reach the buyer?
Give corrections a route and an identity. Explain how a buyer learns that a finding changed, where the corrected version appears and whether the previous record remains accessible. Distinguish an editorial correction from a changed external condition: a vendor raising its price is not the same event as correcting a misread price.
Make it inspectable: retain the identity of the corrected claim and the version it affected. A new publication date should not conceal a correction to evidence on which a customer previously relied.
8. What are the price, renewal, refund and remedy terms?
State the charge, currency, tax treatment where applicable, payment timing, renewal mechanism, cancellation process and remedies. A credit against a future purchase should not be encoded as cash refundable today. A missing price is unresolved; a documented zero price is a known value.
Include the cost of missed delivery in the buyer’s decision. If a time-sensitive brief arrives late, a future service credit may have little value for that decision. The offered remedy, its claim window and any limits should be visible alongside the purchase rather than buried in a disconnected FAQ.
The proposed canvas does not create these obligations. It makes the seller choose and document them. Buyers can then compare actual offers, including offers that provide a one-time answer without a continuing update service.
9. Which text is authoritative when records disagree?
Link the structured record to the exact human-readable terms, identify who issued it and state what governs if versions disagree. Include an effective date, version identifier and route for questions or disputes. Without that rule, an old export can look as authoritative as a revised contract.
ODRL provides concepts for expressing actions, permissions, duties and constraints. It is useful vocabulary for an implementation team, but selecting words from a standard does not settle the underlying bargain. The published canvas uses a small DN record format and does not claim standards conformance.
DCAT 3 distinguishes dataset descriptions and version relationships. That supports a practical discipline: record both the continuing asset identity and the particular version delivered. Do not overwrite a version-specific licence reference with a generic link to whatever terms happen to be current.
Test the representation with purchase-changing questions. Can an agent find the renewal date, permitted recipients and applicable version? Does the output say unknown when a permission is missing? Can a human trace each answer to a clause? Use adversarial cases in which two records differ only in one consequential condition.
Do not accept a successful parser run as proof of faithful interpretation. The downstream system may read every field and still omit an exception when answering. A practical acceptance test compares the actual answer and cited version against the intended meaning of the terms.
Methodology, limitations and corrections
DN used the linked primary specifications to distinguish policy expression, asset identity and versioning, then designed a nine-part commercial drafting canvas. The example is fictional and does not describe an available DN product or promise. The browser tool counts declared states and missing text. It makes no network request and does not assess contract wording. A finished transaction requires the actual reviewed offer and agreement, with specialist advice where material rights or obligations remain uncertain.
Turn the canvas into a concrete purchasing decision
Complete one canvas per offer and intended use. Each of the nine groups has a status and a text field. Mark a group documented only when the relevant terms are actually stated, and add clause references. Explicitly writing “no continuing update service is included” can be a documented term; leaving the service field empty cannot.
The tool counts documented groups, unresolved groups and contradictions. It also lists empty text fields. It exports the inputs and findings so a procurement owner can return with a precise question. Completion measures the record’s coverage, not contract quality, legal enforceability or supplier reliability.
Try a hypothetical fixed-scope brief: a named buyer, one dated PDF and structured record, internal analysis permission, no redistribution of source material, no automatic renewal, and a stated correction route. The unresolved question might be whether an external contractor can process the file. That is one bounded issue to resolve before purchase, not a reason to invent a broad enterprise licence.
For sellers, the commercial benefit is a smaller negotiation gap. A prospect can see which use is included and which needs a separate quote. For buyers, it is a clearer rejection rule: if a required use or remedy remains unresolved, hold that purchase or change the scope.
Keep the executed agreement and the reviewed canvas together. When terms change, create a new record and review affected obligations. A machine-readable commercial offer becomes useful when it helps the buyer make the right commitment and preserve the evidence for that decision.
For corrections, identify the claim, source and date when contacting Decentralised News. A new record-creation timestamp does not refresh its underlying evidence.
Change log: version1.0.0, 2026-11-11. Initial documentation review, original instrument and labelled examples.
Frequently asked questions
Is this DN’s new licence?
No. It is an educational drafting canvas. It does not amend DN’s existing terms, set a product price, grant research rights or promise a service.
Does JSON make a licence enforceable?
The format alone does not establish agreement, authority or legal effect. Keep a reference to the applicable terms, the parties and the relevant version, and obtain the actual agreement.
Can a licence allow retrieval without allowing model training?
A proposed offer can specify different permissions for different activities where the licensor has the relevant rights. The exact wording, scope and applicable law determine the result; the canvas records the offered distinction.
What does a complete canvas mean?
All nine groups have been marked documented and contain text. That is a coverage check on user-entered material, not independent verification or legal approval.
Primary sources
- W3C ODRL Information Model 2.2. Policy expression using permissions, prohibitions, duties and constraints. Reviewed 2026-11-11.
- W3C Data Catalog Vocabulary, version 3. Dataset metadata, version identities and relationships. Reviewed 2026-11-11.
- Creative Commons Attribution 4.0 legal code. Example of an explicit rights grant, conditions and scope limitations. Reviewed 2026-11-11.
Turn the finding into a decision
Use the DN Research Desk to discuss a single research purchasing question with a defined scope and agreed quote. Bring the offer, intended use and unresolved term group. This canvas can help frame the request; any licence, delivery commitment, price or remedy must come from the actual agreed terms.
Related DN research
Disclosure: Analysis and instrument results are editorial. Where labelled affiliate links appear, DN may receive compensation when you sign up or buy through them. Commercial relationships do not determine findings. No tested integration or guaranteed outcome is implied.
