Transactable
Introducing Copyright 2.0

Provenance for the AI Era

Every creator, every work, every system — trusted origins for every digital asset. Drag & drop to make your work instantly transactable.

Activate Copyright 2.0See how it works
Scroll
How the system works

Two sides. One shared record.

Creators need a way to say what they made and how it may be used. AI companies and autonomous agents need a way to find out — in milliseconds, at machine scale, with evidence they can keep. Today neither side can reach the other. Transactable is the layer in between: one registration on the creator side, one answerable query on the machine side, and a signed receipt that both sides can hold up afterwards.

The creator sideRegister · classify · set terms
The layer betweenRegistry · terms hash · gate
The machine sideDiscover · clear · pay
01 — THE PROBLEM

Two sides that can't hear each other

A creator can care intensely about how their work is used and still have no way to say so in a form a machine will act on. A model developer can be entirely willing to license and still have no way to find out who to pay. The failure is not bad faith. It is missing plumbing.

THE CREATOR SIDETHE GAPTHE MACHINE SIDEPhotographer“Not for model training.”✕CrawlerReads robots.txt, finds nothing✕Intent lives in prose, if anywhereSongwriter“Licence it — at a price.”✕Training pipelineNo counterparty, no price, no rail✕An opt-out is a wish, not a grantStudio & its agentsOutput no registry will accept✕Agent at runtimeMust decide in milliseconds✕Nothing either side can keep
Fig. 1 — The gap. Every arrow in this diagram is an intention that exists and cannot travel. The creator's terms are expressed in language no crawler parses; the crawler's willingness to comply meets a file that conveys no rights; and when the exchange does happen, neither party ends up holding anything they could later show a court, an auditor, or each other.
01 · EXPRESSION

Terms aren't machine-readable

A licence written for humans cannot be evaluated at crawl speed. A signal like an opt-out header is legible but conveys nothing — it withholds, it never grants.

02 · DISCOVERY

There is no one to ask

Even a buyer who wants to pay has no address to send the question to. Rights ownership is scattered, unindexed, and for most works never recorded anywhere at all.

03 · EVIDENCE

Nothing survives the transaction

Compliance is asserted after the fact, from logs neither side controls. What's missing is an artefact created at the moment of use that both parties can keep.

02 — THE SHAPE OF THE SYSTEM
Both sides

One registration. One query. One receipt.

The whole system is a loop with six moves — three on each side — turning on a shared record in the middle. Everything that follows in this page is one of these six moves explained properly. Read the diagram once and the rest is detail.

CREATOR SIDETHE LAYER BETWEENMACHINE SIDE01Register the workContent hash · time · claimant02Declare how it was madeOne of five authorship modes03Set the termsUse type × price × conditionsTHE RECORDRegistry entrywork_idauthorship_modeterms_hashVersioned. Never mutated.Every version keeps its hash.04Discover the termsQuery the registry, or a beacon05Clear the useWork + agent both certified06Pay for itPayment references the hashOUTPUTSigned clearance receiptRegistry snapshot hash · terms hash · payment referenceAgent certificate · timestamp · signatureto the creatorto the buyerRegister onceThe creator does this a single time per work.Answer foreverThe record answers every future query automatically.Keep the proofBoth parties hold the same signed artefact.
Fig. 2 — The loop. The asymmetry is the point. The creator side is slow, deliberate and human — it happens once. The machine side is fast, automated and repeated — it happens millions of times against the same record. The layer in the middle exists so that a decision made once by a person can be executed indefinitely by software.
03 — MOVE ONE
Creator side

Register: pin the work to a moment

Registration is not a filing cabinet. It is the act of binding four things together —a work, a time, a claimant, and a signature — so that everything downstream has something stable to point at. The content itself never has to leave the creator's control; what enters the record is a hash of it.

INPUTTHE RECORDWHAT IT MAKES POSSIBLEThe fileImage, audio, text, model outputThe claimantHuman, studio, or an agentThe contextTitle, type, date, descriptionThe basisTools used, process notes, draftsREGISTRY ENTRYwork_idtxbl:work:8f3a…content_hashSHA-256 of the fileclaimantsigned identityregistered_atanchored timestampauthorship_modesee move twoterms_hashsee move threeThe file stays with the creator.Only its fingerprint is recorded.Priority in timeAn independent record of when it existedIdentity that survives the fileFound by fingerprint, not by filenameAn address for termsSomething a buyer's software can query
Fig. 3 — Registration anatomy. Note what is not here: no assertion about who owns what, and no legal conclusion. A registration is a dated, signed claim with a documented basis. Its strength comes from being contemporaneous and tamper-evident, not from anyone having adjudicated it.
Why the hash matters more than the file

Because the record is keyed to a fingerprint of the content, a work can be found again after it has been renamed, re-encoded, stripped of metadata, and passed through six intermediaries. Provenance that lives only in an embedded metadata field is one screenshot away from gone. Provenance that lives in a hash-keyed registry is not.

04 — MOVE TWO
Creator side

Declare how the work was made

Almost every hard question about an AI-era work — is it protectable, must it be labelled, can it be sold, who gets paid — depends on one thing: how much of it a human actually made. So we ask that once, in plain language, and record the answer in a form every downstream system can read.

The taxonomy is deliberately categorical rather than a percentage slider. Asking a creator to say "62% human" invents a precision nobody has. Asking them which of five sentences describes their process is a question they can answer honestly, and one a reviewer can later check against the basis they supplied.

HUMAN AUTHORSHIPMACHINE GENERATIONMODE 01Human Created“I made this entirelymyself, without anyAI tools.”PROTECTABLEYesPRIMARY PATHStatutory copyright+ provenanceMODE 02Human-Led“I directed everything.AI helped me executemy vision.”PROTECTABLELikelyPRIMARY PATHStatutory + AIdisclosureMODE 03Collaborative“Many rounds, backand forth. Our work iswoven together.”PROTECTABLECase by casePRIMARY PATHExaminationrecommends a routeMODE 04AI-Led“AI created this. Iselected and arrangedthe results.”PROTECTABLEUnlikelyPRIMARY PATHRegistry record+ contract termsMODE 05AI Generated“AI made this withoutsignificant creativedirection from me.”PROTECTABLENoPRIMARY PATHRegistry record+ contract termsLeft of centre, copyright law does the work.Right of centre, only infrastructure can.
Fig. 4 — The five authorship modes. The right-hand side of this spectrum is the part the existing system has no answer for. A work with no human author has no copyright to assert — but it still has a maker, a cost, a market, and a buyer who would rather licence it than argue about it. Exclusion there comes from access control and contract rather than from statute.
WHAT IS CAPTURED

A documented basis, per mode

Each mode asks for the evidence appropriate to it — the AI tools used, which elements they touched, roughly how many iteration rounds, the specific model and version, whether output was modified after generation.

WHAT IT IS

An attested classification

The mode is the creator's declaration, recorded with the basis they gave for it and reviewed against that basis. It is strong evidence of what was claimed, when, and on what grounds.

WHAT IT IS NOT

A verdict that a work is AI-free

No system can inspect a finished file and prove no model touched it. We do not claim to. What the record establishes is a contemporaneous, signed, checkable declaration — which is what disputes actually turn on.

05 — MOVE THREE
Creator side

Set terms a machine can act on

"All rights reserved" tells a crawler nothing it can execute. Terms become useful the moment they answer a specific question — may this particular kind of system do this particular thing, and at what price — in a structure software can evaluate without a lawyer in the loop.

Terms are modelled in five layers because five different kinds of statement are being made, and conflating them is how licensing systems go wrong. A standing public licence is not a preference signal. A preference signal is not a grant. A claim someone made on a third-party site is not the same as terms the rights holder set. And an executed licence is different from an offer nobody has accepted.

FIVE LAYERS OF TERMSTHE OFFER MATRIX — EXAMPLEAInstrumentsStanding public licences: CC, CC0, Public Domain Mark,a custom EULA. Binding, but free and already granted.BSignalsPreferences and opt-outs — robots directives, TDMreservations, crawl policies. Mutable. Never a grant.CAssertionsWho claimed what, where, and when — including claimsmade elsewhere that contradict each other.DOffersPriced, payable, versioned terms — the matrix at right.This is the layer that gets hashed.EGrantsLicences actually executed — each one an append-onlyreceipt binding terms, payment and agent identity.USE TYPEPERMISSIONPAYMENTsearchallowedfreeai-indexallowedattributionai-inputconditionalper-useai-trainconditionalpurchasesummarizeallowedattributiondisplayconditionalper-useagent-actionconditionalper-usederivativeprohibited—Eight use types, seven payment types. Each cell is separately transactable.
Fig. 5 — Terms, in layers. The eight use types exist because each one is already priced separately somewhere in the market. Indexing is not training. Retrieval at inference time is not training. An agent acting on a work is not a person reading it. Collapsing them into a single yes/no is what makes existing licences unusable to machines.
// the shape of a terms document{"terms_version": "2026-07-11T00:00:00Z/3","work_id":      "txbl:work:8f3a…","instruments":  [{ "spdx": "CC-BY-NC-4.0", "url": "https://creativecommons.org/…" }],"offers": {"ai-train": {"permission": "conditional","payment":    { "type": "purchase", "amount": 12.00, "currency": "USD" },"attribution": "credit+link","duration":    { "type": "perpetual" },"user_class":  ["commercial"] } },"guardrails":   ["no-likeness", "no-style-prompting"],"territory":    "worldwide","governing_law": "California"}
The rule that keeps this honest

Never sell what a standing licence already grants for free. If a work is published under CC BY, commercial use is already permitted at no charge — offering to sell it back would be misrepresentation. The system refuses to create that offer and returnsalready_permitted instead. What can legitimately be sold on top is the thing CC never provided: a paid licence for uses the public licence withholds, and a receipt proving the use was cleared.

06 — THE HINGE
Both sides

One number both sides point at

Terms change. Prices change. A dispute two years from now is not about what the terms say today — it is about what they said at the moment of use. So every version of a terms document is canonicalised and hashed, and that hash becomes the thing the price quote, the payment, and the receipt all refer to.

AUTHORED ONCECANONICALISEDREFERENCED EVERYWHERETerms documentVersion 3, authored bythe rights holderterms_version …/3Deterministic bytesRFC 8785 canonical JSON,then SHA-256terms_hash = 4c9e…The price challengeQuotes a price for exactly this versionThe paymentPaying against the hash is acceptanceThe receiptBinds the terms that were actually liveVersions are never mutated. A receipt issued against version 3 stays valid evidence of version 3 forever — even after version 4 exists.
Fig. 6 — The terms hash. This is the smallest and most important object in the system. It is what turns "we had an agreement" into "we had thisagreement", and it is why a price quote, a payment and a receipt can be independently verified as describing the same deal.
07 — MOVE FOUR
Machine side

Discovery: make the answer findable

A registry nobody can find is a filing cabinet again. So the record does not wait to be queried — it broadcasts pointers back to itself in every format the machine side already reads, and publishes an explicit rule for what wins when two of them disagree.

Each pointer is generated from the same underlying record, so they cannot drift apart. And because one of them travels inside the file itself, a work that is downloaded, re-uploaded and passed along still carries the address of its own terms.

SOURCE OF TRUTHPOINTERS EMITTEDWHAT WINS ON CONFLICTRegistry recordAuthoritative for everyregistered work.GET /works/{id}/termsLicence file at the domainPer-work terms + a robots.txt pointerHTTP headers & robots directivesConservative preferences, derived from offersWell-known TDM policyThe standard path a scraper already checksStructured page markupLicence + “where to acquire” on the work’s pageAssertion inside the fileTravels with the asset through redistributionHIGHEST AUTHORITY1 · The registry record2 · Assertion embedded in the file3 · HTTP header4 · robots directives5 · Page markupWEAKESTThe beacons are pointers. The record is the answer. Publishing the precedence rule is what makes a conflict resolvable instead of arguable.
Fig. 7 — Discovery and precedence. Most provenance schemes fail at the same place: metadata gets stripped, and there is no fallback. Here the embedded assertion is a convenience, not the source of truth. Strip it and the work is still findable by its content fingerprint, and the registry still answers.
08 — MOVE FIVE
Machine side

The gate: certify both sides, or don't proceed

This is the load-bearing move. Before a use is cleared, the system checks two things at once: what the registry says about the work, and what a certificate says about the party asking. Neither alone is enough. A work with clean terms accessed by an unidentified crawler is not a cleared use, and a well-behaved certified agent does not get a pass on terms it never read.

CERTIFICATION ONE — THE WORKCERTIFICATION TWO — THE ACTORTHE GATEOUTCOMERegistry attestationSnapshot of what the recordsaid at this instantwork_id · authorship_modeterms_hash · snapshot_hashAgent certificateWho is asking, on whosebehalf, certified for whatagent_id · principalscope · validityClearance gateIs this use type offered?Does a standing licencealready permit it free?Is the agent in scope?Have the conditions been met?All four, or no clearance.ClearedAccess granted for this use,this agent, this version of terms.→ signed clearance receiptRefusedNo offer for this use, agent outof scope, or conditions unmet.→ signed negative receiptA refusal is recorded as carefully as an approval. Proving you were told no is often worth more than proving you were told yes.
Fig. 8 — Dual certification. Because the gate is a condition of access rather than a notice posted afterwards, the party passing through it has actually accepted the terms attached to it. That is what makes the arrangement contractual rather than merely declaratory — and it is why the gate, not the registry, is the commercially load-bearing part of the system.
ON THE WORK SIDE

A snapshot, not a lookup

The gate does not record "we checked the registry." It records the exact state the registry was in, hashed, so the answer can be re-verified later without trusting anyone's log.

ON THE ACTOR SIDE

Agents get identities too

An autonomous agent is a party to the transaction. It carries a certificate naming its principal and its permitted scope — which means an out-of-scope action is refusable at the gate, not litigated afterwards.

THE HARD LIMIT

Strangers who never knock

The gate binds those who pass through it. A party that never touches the gate is outside this system entirely and remains a matter for law, not plumbing. We are direct about that boundary.

09 — MOVE SIX
Machine side

Payment at the speed of the request

Licensing at machine scale cannot route through a sales team. When an agent asks for a use that costs something, the answer is a priced challenge carrying the terms hash — and the agent can settle it and continue, in one round trip, with no human on either end.

The economics only work because the amounts are small and the overhead is near zero. Per-use prices in the cents range are viable when settlement costs a fraction of a cent and requires no account opening, invoice, or reconciliation. That is a recent development, and it is what makes per-use rights markets possible at all.

AGENTRESOURCE + GATESETTLEMENT RAIL01Request a use — “ai-input on work 8f3a”02402 Payment Required · price · terms_hash · terms URLAgent evaluates: in budget? in scope? in policy?03Settle, referencing terms_hash — this is acceptance04Payment confirmedGate re-checks both certifications05200 OK · the resource · and a signed clearance receiptPaying against a hash is what converts an offer into a grant. The agent did not agree to “the terms” — it agreed to version 3, byte for byte.
Fig. 9 — The clearance handshake. Nothing here is novel as protocol — it is a payment challenge, a settlement, and a retry. What is new is the payload: the challenge carries the identity of the rights being conveyed, so the payment is not just money moving but a licence being formed.
10 — THE OUTPUT
Both sides

The receipt is the product

Everything before this exists to produce one artefact. A clearance receipt is a signed statement, generated at the moment of use, that binds together the work, the terms as they then stood, the payment, and the identity of the party acting. Both sides get the same one. Neither side has to trust the other's records.

IT BINDSTHE ARTEFACTTHREE KINDSThe work, as recordedSnapshot hash of the registry stateThe terms, as they stoodterms_hash of that exact versionThe considerationPayment reference on the railThe party actingAgent certificate and principalThe momentTimestamp, anchored twiceCLEARANCE RECEIPTsnapshot_hashterms_hashuse_typepayment_refagent_certissued_atsignatureAnchored to two independentsubstrates, so neither can bequietly rewritten later.PositiveA use was cleared and paid for.Evidence of a licence, held by both sides.NegativeA use was requested and refused.Evidence the buyer asked and was told no.DeletionA work was withdrawn or erased.Evidence the removal actually happened.A receipt is not a claim about who was right. It is a tamper-evident record of what the system said and did, at a moment, to a named party.
Fig. 10 — Receipt anatomy. The negative receipt is the underrated one. Most compliance disputes turn on what a party knew and when — and a signed refusal, timestamped, is a far cleaner answer than a server log either side could have edited.
11 — THE PAYOFF
Both sides

Register once, project everywhere

A creator should not have to describe the same work five times to five institutions in five formats. The record holds the facts once; every external regime is aprojection of that record in whatever shape that regime demands — and when the underlying facts change, every projection changes with them.

ENTERED ONCEPROJECTED OUTWARDOne registrationThe work, its fingerprint,its authorship mode, thebasis, and its terms.the single sourceCopyright office filingWhere the work is registrable, the record already holds every field the application asks for.Transparency disclosureAuthorship mode supplies the substantiated basis behind an AI-content marking obligation.Platform AI labelsA registered Human Created work becomes appeal evidence against an incorrect automatic label.Crawl and mining signalsMachine-readable preferences, derived conservatively from the offer matrix rather than hand-written.Embedded content credentialsAn assertion carried inside the file, pointing back to the record it came from.
Fig. 11 — Projection. This is the inversion. The conventional model treats the copyright office as the system of record and everything else as a copy. Here the registry is the system of record and the office filing is one projection among several — an important one, but not the origin.
12 — THE HONEST PART
Both sides

What this proves — and what it doesn't

Trust infrastructure that overstates itself is worse than none, because the first serious challenge collapses it. So here is the boundary, drawn deliberately.The system is excellent evidence of what happened. It is not an adjudicator of who was right.

WHAT IT ESTABLISHESWHAT IT DOES NOT✓That a claim existed at a given momentDated, signed, and anchored so it cannot be back-written.✓Exactly what the terms said thenNot today’s terms — the version that was live at the time of use.✓That a named party accepted themCertified identity, in scope, having passed the gate.✓That consideration changed handsA settlement reference that can be checked independently.✓On what basis a mode was declaredTools, process, model and version — kept with the declaration.✕That the claimant truly owns the workA registry records assertions. It cannot verify title from the outside.✕That a work is genuinely AI-freeA mode is an attested classification with a documented basis, not a detector.✕That any party is compliant with a regulationThose duties sit with providers and deployers. We supply audit-grade evidence.✕That a particular use was lawfulExceptions and fair use are for a court. We do not adjudicate them.✕Anything about a party who never knockedThe gate binds those who pass through it. Strangers remain a matter for law.
Fig. 12 — The boundary. Read the right column as a feature. Every one of those limits is a place where a competing system would be tempted to overclaim, and where overclaiming is exactly what would make the evidence worthless when it mattered.
The claim we are careful about

A receipt proves what the registry said and what the parties did. It does not prove that the registry was right about the world. Those are different things, and a system that blurs them is selling certainty it cannot deliver. Ours is designed to be court-ready evidence — not a substitute for the court.

13 — TIMING

Why this becomes load-bearing now

Two things are converging. Regulation is beginning to demand that AI-generated content be marked and that its provenance be substantiable. And the open web is shifting from crawl-by-default to block-by-default, which turns access into something that must be negotiated rather than assumed.

2 AUG 2026Transparency duties applyAI content must bemarked and disclosed15 SEP 2026Block-by-default crawlingAccess becomes negotiated, not assumed2 DEC 2026Grace period endsSystems already on the marketlose their transition allowanceDEC 2027High-risk obligations beginEvidence requirements deepenAUG 2028Remaining phase-in
Fig. 13 — The forcing dates. None of these regimes tells anyone how to produce the evidence they require. That gap between an obligation and a mechanism is exactly where infrastructure gets adopted.
  • ✓The obligation lands on the AI side. Providers and deployers must mark and substantiate. They need a source of truth they did not author themselves.
  • ✓The leverage lands on the creator side. When crawling is blocked by default, a creator's terms stop being a request and start being a condition of entry.
  • ✓The rail finally exists. Sub-cent machine-to-machine settlement makes per-use licensing arithmetically possible for the first time.

Transactable — provenance, terms, clearance and compensation for works made by humans, models, and agents.

This page describes system architecture. Nothing here is legal advice, and nothing here asserts that any party is compliant with any regulation — those duties sit with the providers and deployers of AI systems.

AI-Assisted Analysis

Get instant, proprietary analysis that verifies your work and streamlines your registration process.

International Evidence

Generate robust, globally-recognized evidence of creation for protection that extends beyond US borders.

Monetize the AI Economy

Position your work now to ensure you’re compensated when AI models train on your data.

Copyright 2.0 Platform Overview
The Wrapper Layer

The Future of IP Protection Has Arrived

Copyright 2.0 is the digital rights layer you need.

Your creative work deserves more than a static registration. Transactable Copyright 2.0 is the essential, technology-forward layer built on the foundation of the traditional international copyright regime.

Beyond protection. Beyond borders.

What changes? Everything. We combine blockchain provenance, real-time AI assessments, and NFT transactability to turn your IP into a liquid, licensable, automatically monetizing asset.

The global copyright framework is your shield. Our platform is your engine — we overlay the existing international regime with next-generation technology to create entirely new opportunities for creators.

Upgrade to Copyright 2.0See the Difference
Architecture · Copyright 2.0

The Protocol for Digital Ownership

Infographic

Copyright 1.0 vs. 2.0

The evolution from traditional copyright protection to the modern, AI-enabled system

Copyright 1.0

Traditional System

  • Proof of Creation:

    Paper records, optional registration

  • Monetization:

    Licensing, sales after the fact

  • AI Content:

    Unregisterable / Legally Unclear

  • Process:

    Slow, complex forms, legal fees

Copyright 2.0

Modern System (Transactable)

  • Proof of Creation:

    Instant, Immutable NFT Provenance

  • Monetization:

    Passive Income from AI Training Fees

  • AI Content:

    Future-Proofing for eventual USCO eligibility

  • Process:

    Simple drag-and-drop, on-ramp to official USCO registration

Choose Your Copyright 2.0 Path

Different creators have different needs. Find the approach that fits your goals.

AI Artists

Claim Your Seat at the Table Before the Law Changes

The US Copyright Office is still evaluating AI-generated works. By registering today with Copyright 2.0, you establish verifiable proof of creation and ownership that will be invaluable when regulations crystallize. Your blockchain certificate creates an immutable timeline—positioning you ahead of legal clarity.

Proof Now. Protection Later.

  • Establish cryptographic provenance today
  • Complete transaction history from day one
  • Ready for US Copyright Office registration when rules change
  • Monetize through AI training licensing now
Future-Proof Your AI Art

Traditional Creators & Agencies

The Dual Layer of Protection

Transactable provides the instant, international digital evidence layer (the Copyright 2.0 Wrapper) that backs up and future-proofs your claim. For applications In the US, you can also obtain registration with the US Copyright Office which provides important additional rights in the United States, including the right to sue for copyright infringement and to recover damages and attorney fees.

Instant Provenance. Future Security.

  • Instant blockchain provenance certificate
  • Streamlined path to official US Copyright Office registration
  • Protection for traditional and AI-assisted works
  • Peace of mind with dual-layer security
Secure My Copyright

Digital Natives & NFT Enthusiasts

Turn IP into Liquid Capital

Stop viewing copyright as a stale legal certificate. Copyright 2.0 is Transactable. Our Wrapper Layer takes your static digital file and converts it into a liquid, verifiable IP asset (the NFT). This instantly makes your work transactable for sales, licensing, and AI training fees, future-proofing your revenue stream beyond traditional markets.

From File to Fundable Asset.

  • NFT certificate with instant liquidity
  • Programmable royalties through smart contracts
  • Passive income from AI training licensing
  • Trade IP assets on secondary markets
Make My Work Transactable

Developers & Tech-Savvy Creators

The Protocol for Digital Ownership

We're not just a platform; we are the new protocol for digital asset ownership. Copyright 2.0 is this standard. Transactable provides the foundational Wrapper Layer that links a file to a verifiable entity, making the IP inherently transactable across any blockchain or legal system. This architectural foundation future-proofs digital ownership against technological shifts.

The IP Layer for the Next Internet.

  • Cryptographic proof via file hash verification
  • Smart contract integration for licensing
  • API access for programmatic registration
  • Cross-chain compatibility for NFT certificates
Build on Copyright 2.0

Agencies & Professionals

Automation for IP Compliance

Managing IP protection shouldn't require a full-time lawyer. Our Copyright 2.0 Wrapper streamlines the entire process, providing instant NFT proof and an easy, pre-formatted path to US Copyright Office registration. It eliminates the friction of Copyright 1.0, making compliance effortless and future-proofing your entire project catalog.

Compliance Simplified. Protection Maximized.

  • Bulk upload for multiple works at once
  • Automated US Copyright Office filing preparation
  • Dashboard for managing client portfolios
  • White-label certificates for client delivery
Streamline My IP Protection

AI Agents & Developers

Participate and Profit in the AI Economy

You deserve to be compensated for work. As Artificial Intelligence (AI) models increasingly train on existing data, content creators must be ready to get paid for their contributions. It's inevitable that AI companies will need to pay creators when they use their works—and you need to be in line to receive that compensation. Our platform is the essential link between creators and the evolving AI ecosystem. We provide a path for you to not only enforce protections on your copyrighted digital works but also to easily monetize your data when it's used for AI training.

The Transactable Path to Ethical AI.

  • Verified creator compensation on-chain
  • Reduce litigation risk with compliant data
  • API access to licensed training datasets
  • Transparent provenance for every work
Access Ethical Training Data

How It Works

We combine AI analysis, blockchain provenance, and government registration into a 5-minute process. Drop your file, answer a few questions, and your IP Agent handles the rest—processing your work, generating your Certificate of Provenance, minting your NFT, and preparing your official registration.

What you do

You can launch your IP Agent in two easy steps.

1

Drop file

Upload your file

Drop your digital file (code, document, schematic) on our uploader.

2

Answer a few questions

Answer a few questions

Answer a few questions about your work. Process takes less than a minute.

What we do

We've condensed months of legal processes into minutes.

1
Initialize root agent

Initialize Agent

After you drop your file on our uploader, we initialize your IP Agent to start the registration process.

Your profile information, digital file, and answers to our questions are all we need to process your work.

2
Process file

Process File

Your IP Agent submits your file to the File Processing Component to perform the following operations:

  • Store the file in our secure file storage
  • Create a cryptographic hash of the file
  • Programmatically determine the file type
  • Create a processed file based on the file type
  • Provide processed file for LLM analysis
3
AI Analysis

AI Analysis

Your IP Agent then submits the processed file to the following sub-agents:

  • Describer Agent generates a description of the file contents
  • Assessment Agent leverages vector store to generate a preliminary copyright assessment based on jurisdiction
  • Valuation Agent generates a valuation of the work for access purposes
  • Monetization Agent generates a monetization plan for the work based on jurisdiction
4
Certificate generation

Certificate Generation

A Certificate of Provenance is generated by the Certificate Generator with the following information:

  • Pertinent copyright information about the work
  • AI‑specific information about the work
  • QR code of the cryptographic hash of the file for verification
  • QR code of the smart contract address for payments and transactions
5
Blockchain operations

Blockchain Operations

The Blockchain Operations Component anchors the Certificate of Provenance to the blockchain for tamper‑evident records. It is responsible for:

  • Creating your smart wallet (part of profile onboarding)
  • Minting the NFT representing Certificate of Provenance
  • Anchoring the Certificate of Provenance to the blockchain
  • Storing the NFT on a decentralized storage system (IPFS)
6
Dashboard

Dashboard

The Dashboard is where you can manage assets from one place, including:

  • Manage your assets
  • Download your Certificates of Provenance
  • View your transactions
  • Monitor licensing opportunities

Opportunities

With your work registered, your IP Agent is ready to help you monetize your work.

Digital signature

Government Registration

If your jurisdiction has a government registration process, your IP Agent can facilitate the process. Countries with government registration include:

  • United States
  • Canada
  • China
  • India
Payments

AI Training

Your IP Agent can accept requests from LLMs to train on your work. If you check the "Available for AI Training" box on your registration, your IP Agent will:

  • Accept requests from LLMs to train on your work
  • Verify that the access fee has been deposited to your smart wallet
  • Provide the LLM with a time-scoped access token to ingest your work
Transactions

Transactions

With your work AI and blockchain enabled, the possibilities are endless. Your IP Agent makes your work infinitely transactable:

  • License your work to others
  • Sell your work
  • Engage in other transactions through smart contracts
Projects

Portfolio

Your IP Agent is ready to help you manage your assets and transactions, and build your intellectual property portfolio.

Building the Trust Layer for the AI Economy

The AI economy runs on other people's work — and today, neither side can prove anything. Creators can't show what's theirs or get paid when models use it. AI companies can't show their agents respect anyone's rights. We're extending the Transactable registry into the infrastructure that fixes both: provable ownership on one side, provable behavior on the other, and a licensing market between them.

Clearance Receipts

Every rights check produces a cryptographically signed receipt — including negative results ("we checked; it was not registered"). Receipts are anchored to a public, append-only log that anyone can verify independently, forever.

Register, Vault & Earn

Anchor your work with a low-cost registration, add secure custody when you want it, and opt in to Earn to license your catalog for AI training — on your terms, at your price floor, with a transparent payout ledger.

AI Agents That Verify First

We train Cleared-Use models whose commitment is built into their weights: verify registration and license terms before using any work — and decline with a licensing path when clearance cannot be verified.

Behavior That Is Measured, Not Promised

Every trained commitment is evaluated on held-out scenarios it never saw in training, against a published pass bar — our first Cleared-Use model held its commitment on 97.8% of out-of-domain tests.

Provenance You Can Trace

The commitment itself is a registered, hash-anchored asset on our own registry, and every training run — base model, data fingerprint, evaluation results — is recorded against it and publicly queryable.

A Market Where Both Sides Win

Verified agents license cleared work through automated payment rails; creators get paid and agent operators get audit-ready proof of compliance. The registry sits in the middle, keeping everyone honest.

Register once. Prove it forever. Get paid when AI uses it — and know the agent on the other side was built to ask first.

Bulk Registration

Have a large number of works to register? Our bulk workflows are built for power users, publishers, and companies representing authors. Choose the method that fits your pipeline.

Web

User‑Friendly Web Interface

Bulk upload via web interface

Upload many works using our guided bulk flow. Drag and drop, validate entries, and submit securely with payment in one streamlined experience.

Contact our team
Zip

Zip + CSV Manifest

Zip file with CSV manifest

Prefer offline prep? Drop a single .zip containing all files and a CSV manifest. We'll parse, validate, and process each row asynchronously with status tracking.

Contact our team
API

API‑Only Integration

Integrate directly from your systems using an API key. Create, track, and finalize registrations programmatically.

curl -X POST https://api.transactable.ai/v1/bulk/registrations \
-H 'Authorization: Bearer <API_KEY>' \
-H 'Content-Type: application/json' \
-d '{
"callback_url": "https://yourapp.com/webhooks/transactable",
"items": [
{
"title": "Work 001",
"file_url": "https://cdn.yourapp.com/works/001.pdf",
"jurisdiction": "US",
"ai_training": false
},
{
"title": "Work 002",
"file_url": "s3://bucket/works/002.wav",
"jurisdiction": "CA",
"ai_training": true
}
]
}'

Tip: Use webhooks to receive processing updates per item.

Request API Access
Activate · Copyright 2.0

Ready to Upgrade to Copyright 2.0?

Join the future of IP protection — instant provenance, official registration, and passive income from AI training.

Launch My Copyright 2.0 Protection→Talk to Our Team