On this page
Start with the deliverable
Rightmark's offer Rightmark, (opens in a new tab) and buyer process Rightmark, (opens in a new tab) explain those steps.
Defined.ai describes a marketplace, custom collection and evaluation services in its March 25, 2026 guide Defined.ai, (opens in a new tab). That gives buyers a useful choice: start with a private source and its licensing scope, or evaluate a dataset or service proposal against the task. Owners should confirm the current contribution route and the agreement they would sign.
The important comparison is the same asset, intended use and accepted deliverable. An owner's percentage alone cannot tell an AI buyer what the project will cost or whether the data will work.
Bring the asset or the AI task. Owners can start with Rightmark's appraisal, which needs answers rather than raw files. Buyers can describe the data they need Rightmark, (opens in a new tab), including the intended task, exclusions and minimum useful sample.
Rightmark publishes this comparison. Research was checked October 22, 2026, in New York. Defined.ai's marketplace claims use an archived official article bearing the publication date March 25, 2026. The archive was captured May 16, 2026; its live article refresh was unavailable. The buyer license carries its own effective date. Neither source establishes an owner's current partnership terms.
Rightmark vs Defined.ai at a glance
Compare the route you would actually enter. An owner partnership agreement and a buyer's dataset license answer different questions.
| Decision | Rightmark | Defined.ai |
|---|---|---|
| Owner entry point | Private-asset appraisal and licensing scope | Confirm the current data-contribution route |
| Buyer starting point | Written private-data requirement | Marketplace or commissioned work |
| Commercial comparison | Owner proceeds and total buyer quote | Partner terms and buyer quote |
| Evidence to request | Route ownership pack, data card and sample | Dataset-specific sample and scope |
| Permission check | The agreement for the selected route | Applicable buyer or partner agreement |
The table summarizes the sections below. Neither a marketplace listing nor an appraisal establishes availability, agreed rights or a completed transaction.
What are you licensing or commissioning?
A useful archive and a finished training dataset are different deliverables. An owner might control the source material but still need to separate excluded content. A buyer might need annotations or newly collected examples that the existing material cannot provide.
Rightmark describes private sourcing against a buyer's specification, with route-specific ownership evidence and graded batches from the owner's systems. That is a concrete reason to evaluate it when source context, permissions and the boundary around a private collection matter. Confirm what can actually be sourced for the requirement. Buyer workflow Rightmark, (opens in a new tab).
Defined.ai's dated guide describes marketplace datasets, custom collection and evaluation support. Official guide, March 25, 2026 Defined.ai, (opens in a new tab).
| Scope question | Rightmark proposal | Defined.ai proposal |
|---|---|---|
| What already exists? | Identify the owner-controlled material proposed for licensing | Identify the proposed dataset or source material |
| What must be added? | Specify preparation or missing evidence | Separate existing data from any commissioned work |
| What counts as delivery? | Agree the batch, metadata and ownership evidence | Agree the dataset or service deliverable and evidence |
The decision: Rightmark is a strong starting point for a private-source licensing requirement. Defined.ai merits consideration when the buying process includes a dataset search or a separately scoped data service. Treat these as choices about the work, not guarantees about either provider's inventory.
Ask both: “Which part of this proposal licenses existing records, which part creates or labels data, and what will prove the delivered material fits our task?”
Are you contributing an asset or providing a service?
An owner licensing recordings and a supplier hired to collect recordings have different obligations. The first is granting rights over an existing asset. The second is accepting responsibility for a project. Comparing their potential income without that distinction produces a misleading offer comparison.
Ask Defined.ai to identify the current agreement for your proposed contribution. If you control an existing archive, ask what can be licensed, what evidence is required and what triggers payment. If you are offering collection or annotation work, request a separate statement of work. Do not infer the terms of either arrangement from its buyer-facing dataset license.
For an owner evaluating Rightmark, the starting point is the private asset and the authority to license it. Its initial appraisal does not require source files or passwords. Later reports and specifically approved samples are a separate step, under the documented review process. Rightmark Trust Rightmark, (opens in a new tab).
| Owner question | Evidence to put in either proposal |
|---|---|
| What do we control? | Asset scope, source systems and authority records |
| What must stay out? | Client-owned material, restricted content and other exclusions |
| What are we being paid for? | A defined license or a separately specified service |
| What happens before sharing? | Recipient, purpose, approved sample and applicable agreement |
The decision: Choose the program that pays for the contribution you intend to make. Rightmark gives private owners a practical assessment and licensing path. A professional data-service supplier should evaluate the relevant project-work program separately from an owner's license.
Ask both: “Which agreement am I entering, what am I contributing, and what event creates a payment obligation?”
Compare owner proceeds and buyer total on the same scope
Rightmark's published standard model is clear: the buyer pays its fee on top of the owner's agreed payout. Its homepage also states that buyers cover extraction, scrubbing and transfer. Premium and bespoke arrangements have a different share agreed in writing. These are published offer terms, not proof that every Rightmark transaction yields the highest owner payout. Fee explanation Rightmark, (opens in a new tab).
For owners, that separation is valuable. It lets you ask whether a competing offer changes the fee allocation, the rights or the work required. For buyers, it means requesting the complete acquisition cost. Keeping more of an owner's agreed payout does not establish a lower buyer bill.
Request Defined.ai's actual commercial proposal for your contribution. Ask for the share basis, deductions, payment trigger and retained rights; this comparison does not establish a standard owner percentage.
| Compare in writing | Owner's question | AI buyer's question |
|---|---|---|
| Commercial basis | What amount is my share calculated from? | What is the full quote for the agreed scope? |
| Preparation | Which costs can reduce my proceeds? | Which extraction, labeling and review work is included? |
| Rights | What term, exclusivity and uses am I granting? | Which uses, outputs and reuse are included? |
| Payment | What event triggers payment? | What delivery or acceptance event triggers payment? |
| Differences | What can change the offer? | How are scope changes or mismatches handled? |
The decision: Rightmark is worth shortlisting for owners who want an explicit standard fee allocation. The better signed offer still depends on net proceeds, granted rights, work and payment conditions together. Use the licensing guide Rightmark, (opens in a new tab) for the underlying terms; ask each provider to answer this same commercial brief.
Ask both: “Show owner proceeds and buyer total separately for this exact scope, including preparation, permitted uses, payment events and exclusions.”
Match the AI task to the actual permission
An internal speech recognizer, a recognizable synthetic voice and a product distributing raw audio are different intended uses. A technically suitable sample does not settle permission for all three.
Defined.ai's published buyer DLA, effective November 30, 2022, grants a non-exclusive license with restrictions on sublicensing and transfer. It permits model commercialization within stated boundaries, excluding disclosure of the data itself or recognizable contributor outputs. Conditional indemnity and liability limitations also apply. Check the actual agreement and exhibits; this buyer DLA does not establish an owner's partnership terms. Published DLA Defined.ai, (opens in a new tab).
Rightmark also requires the actual route to be read. Some code agreements authorize later non-exclusive licenses after signing. Depending on the route, a licensing partner can have sublicensing, own-use or derived-dataset rights. Some code agreements leave the owner's ownership indemnity uncapped. Rightmark publishes these qualifications and says it negotiates caps where buyers allow. Retaining ownership should never be read as retaining an approval over every later license. Route-specific terms Rightmark, (opens in a new tab).
The Data Provenance Initiative found source and dataset-license restrictions can differ in public datasets, without auditing these providers. Research, February 2025 Data Provenance Initiative, (opens in a new tab).
| Check for either provider | Evidence to request |
|---|---|
| Intended task | Training, evaluation, retrieval, display or another named use |
| Source permission | Rights and contributor evidence for the included material |
| Final grant | The current agreement, exhibits and written exceptions |
| Outputs and reuse | Rules for derived assets, recognizable content and onward access |
| End of term | Retention, deletion and surviving obligations |
The decision: Favor the proposal that documents your actual use. A broad “AI training” label is insufficient when the intended outputs or recipients change. Have the people responsible for legal review assess the relevant clauses; this comparison is not legal advice.
Ask both: “Does the signed grant cover our exact training, output and onward-use plan, and which source permissions and exclusions support that answer?”
Test a sample against the delivery you will accept
A small sample can show that a format opens or an annotation looks useful. It cannot prove that the full delivery contains the same mix, has the same permissions or meets the same acceptance criteria.
Rightmark's sample guidance makes that boundary explicit. Its buyer process pairs the data with an ownership pack and a data card describing contents and known gaps. That gives a buyer concrete evidence to request and an owner a clearer description of what will be reviewed. Sample guidance Rightmark, (opens in a new tab), buyer evidence Rightmark, (opens in a new tab).
Defined.ai's guide describes free samples for assessing fit and custom requests when off-the-shelf data is insufficient. Official guide, March 25, 2026 Defined.ai, (opens in a new tab).
Consider this hypothetical example, not a customer result: two audio samples contain the requested language. One matches the required recording channel and annotation task. The other needs additional work or permission. Their equal duration would not make them equivalent purchases. The proposal should expose that difference before either side commits.
| Acceptance decision | Ask both providers to specify |
|---|---|
| Representative sample | Sources, periods, formats and exclusions represented |
| Useful unit | What counts after duplicate, scope and quality checks |
| Final delivery | Manifest, metadata, annotation state and known gaps |
| Mismatch process | Who evaluates a discrepancy and what response is agreed |
For code, use code-specific evidence. Rightmark's Repo Check Rightmark, (opens in a new tab) can support an initial local assessment without uploading source code. It reads available evidence; it does not run the tests or establish ownership, legal clearance or a buyer's acceptance.
The decision: Rightmark's documented ownership and sample workflow is a useful basis for private-data diligence. Apply the same test to every proposal: can the reviewed sample be connected to the exact batch and permission being purchased? Do not assume either provider has a standard refund or replacement policy without written terms.
Ask both: “How will we compare the sample with the delivered batch, and what happens if scope, usable units or rights evidence differ?”
Who carries preparation and follow-through?
The quote needs to identify work as well as data. Someone must prepare evidence, resolve exclusions, approve the sample, receive the delivery and decide whether a mismatch matters. Leaving those roles implicit can make a nominally complete proposal incomplete for your team.
Rightmark publishes an ordered path from written scope through evidence and sample review to a license and delivery. That makes it possible to assign responsibilities against a visible process. Its description remains a process commitment, not evidence of measured staff-time savings or assured results. Buyer workflow Rightmark, (opens in a new tab).
| Responsibility | Record in each proposal |
|---|---|
| Owner evidence | Role responsible for authority, exclusions and permissions |
| Preparation | Included work, required inputs and dependencies |
| Technical review | Evaluation criteria and responsible reviewer |
| Transfer | Approved recipient, access method and revocation process |
| Acceptance | Decision-maker, contract version and mismatch handling |
The decision: Compare the work your team must supply alongside the provider's responsibilities. Rightmark is a clear next conversation for a bounded private-owner license. If you need a broader service engagement, obtain a statement of work that specifies the additional deliverables rather than assuming they are included in a dataset quote.
Ask both: “Who owns each step, which dependencies fall on our team, and what remains outside this proposal?”
What makes Rightmark worth shortlisting?
Three parts of its published process give owners and buyers a concrete evaluation path:
- Explain the economics before comparing outcomes. The standard fee allocation separates the owner's agreed payout from the buyer-paid fee. Confirm the chosen route and full quote. Owner offer Rightmark, (opens in a new tab).
- Start with bounded evidence. The appraisal needs no source files; code owners can use a local readiness report. Broader sample access is a later, controlled step with its own purpose and limits. Trust Rightmark, (opens in a new tab), Repo Check Rightmark, (opens in a new tab).
- Connect the buyer requirement to the source. A written specification, route ownership pack and sample review give the buyer checkable questions about scope and rights. They do not guarantee model improvement. Buyer process Rightmark, (opens in a new tab).
These are positive reasons to evaluate Rightmark. They are not claims that no other provider can offer similar processes.
When is Defined.ai the better fit?
Consider Defined.ai for marketplace procurement or custom collection and evaluation. Its dated marketplace guide Defined.ai, (opens in a new tab) supports that buyer-facing fit. Owners should confirm whether their asset fits a current contribution route and compare the proposed agreement.
Make the selection against the actual proposal. Confirm the dataset, preparation, permissions and economics relevant to your case rather than treating a company-wide description as a commitment to fulfill it.
When is Rightmark the better fit?
Choose Rightmark as a starting point when your owner-side priority is to define a private asset, establish authority and understand licensing economics before broader sharing. It provides an initial appraisal path and publishes the contractual questions that matter after that assessment.
For AI buyers, shortlist Rightmark when source-specific material and a documented ownership and sample path are central to the requirement. Send a bounded specification, then evaluate the response against the same criteria used for other proposals. Confirm current availability for the data family; Rightmark does not currently accept health data. Buyer scope and eligibility Rightmark, (opens in a new tab).
Consider a different path if a properly licensed public dataset already answers the task, or pause the transaction if essential permissions remain unresolved. A provider choice cannot resolve rights the owner cannot establish.
Take one comparable brief to both providers
Use this blank register to keep the owner grant and buyer purchase aligned. Keep completed versions private; it is a decision aid, not an upload form or a legal template.
| Decision | Owner-approved scope | Rightmark proposal | Defined.ai proposal | Evidence or open item |
|---|---|---|---|---|
| Existing material or new collection | [Fill in] | [Fill in] | [Fill in] | [Fill in] |
| Exact AI task and intended outputs | [Fill in] | [Fill in] | [Fill in] | [Fill in] |
| Sources, periods and exclusions | [Fill in] | [Fill in] | [Fill in] | [Fill in] |
| Source and contributor permissions | [Fill in] | [Fill in] | [Fill in] | [Fill in] |
| Final grant, term and onward use | [Fill in] | [Fill in] | [Fill in] | [Fill in] |
| Sample and acceptance criteria | [Fill in] | [Fill in] | [Fill in] | [Fill in] |
| Format, metadata and annotations | [Fill in] | [Fill in] | [Fill in] | [Fill in] |
| Owner proceeds and fee basis | [Fill in] | [Fill in] | [Fill in] | [Fill in] |
| Buyer total and included work | [Fill in] | [Fill in] | [Fill in] | [Fill in] |
| Payment and mismatch events | [Fill in] | [Fill in] | [Fill in] | [Fill in] |
| Responsible role and agreement version | [Fill in] | [Fill in] | [Fill in] | [Fill in] |
Rightmark's buyer request brief Rightmark, (opens in a new tab) provides a fuller specification template. Complete the parts you know and identify the rest as questions. A clear unanswered item is more useful than an assumed term.
Owners: Start an appraisal to describe the asset without sending raw files. AI dataset buyers: Send a scoped data request Rightmark, (opens in a new tab) and use the sample, rights and full-quote questions above to evaluate the response.
Sources
- Rightmark: Buyer process (opens in a new tab)
- Rightmark: Owner offer (opens in a new tab)
- Defined.ai: The AI Marketplace: Understanding the Model Behind Enterprise AI and How It’s Driving Change (opens in a new tab)
- Rightmark: Trust (opens in a new tab)
- Rightmark: licensing guide (opens in a new tab)
- Defined.ai: Data License Agreement (opens in a new tab)
- Data Provenance Initiative: Bridging the Data Provenance Gap Across Text, Speech and Video (opens in a new tab)
- Rightmark: Sample guidance (opens in a new tab)
- Rightmark: Repo Check (opens in a new tab)
- Rightmark: buyer request brief (opens in a new tab)