A provider holds your key
Payroll, claims, KYC, card embossing. Somebody outside the company can decrypt the data because that was the only way to let them work on it, and the same key opens every delivery they ever received.
ANKASecure© replaces the long-lived private key or the cleartext file an outsourced process usually receives with a capability grant. Five objects describe each relationship: an exchange context for the counterparty and the keys in scope, the cryptographic actor that invokes, authenticated through workload identity federation, the capability granted, the key protected by the owner’s HSM or KMS, and a constraint policy that enforces a validity window, a usage cap, a per-minute rate limit and single use. The existing channel (SFTP, managed file transfer, APIs or messaging) stays in place. Every invocation is recorded in a signed audit trail forwardable to a SIEM, revoking the grant refuses the next invocation without re-encrypting data, and the key and algorithm behind a grant can change, including to post-quantum algorithms, without any change on the third party’s side.
Sharing data with a third party usually means handing over a long-lived key, or decrypting the data and sending it in the clear. Either way, control ends at the moment of sharing. ANKASecure© replaces both with a capability grant: the counterparty invokes only the operations it was granted, within enforced constraints, every use is recorded in your domain, and every future operation is refused the moment you revoke it. Your counterparty keeps receiving data over the channel you already use, whether SFTP, an API, messaging or an MFT platform.
Share the information. Keep the authority.
Is this you
Payroll, claims, KYC, card embossing. Somebody outside the company can decrypt the data because that was the only way to let them work on it, and the same key opens every delivery they ever received.
A CSV over SFTP to a collections BPO, a JSON payload over TLS to a fintech. It arrives readable on their servers, and you have no record of which records were opened, or when.
A contract terminates, a provider has an incident, or a supervisor asks who can decrypt what and how you prove it. With a shared key or a cleartext file, the answer is a project, not a decision.
Who signs this: A CISO or head of third-party risk who owns the provider relationship, and a data protection officer who has to answer for what the provider can see. If revoking a counterparty means re-encrypting data or rebuilding a pipeline, this is the conversation.
Why existing approaches fall short
Transport protection gets information to the third party safely, and encryption keeps it unreadable in transit and at rest. Neither decides when, under which conditions and how often the third party may open it.
Who holds the authority to decrypt?
Can access be revoked without the third party’s cooperation?
Can use be limited (once, a time window, a cap)?
Is there evidence of use in your domain?
Does exposure grow over time?
Can the algorithm change without a bilateral project?
Is the relationship with the third party modeled?
Do you control when and under which conditions cleartext is obtained?
Most third-party exchanges are recurring. When a key is handed over once and reused for every delivery that follows, after a year the provider holds 24 deliveries under the same key, and that key opens every one of them, at any time, as many times as it wants, along with anything else ever encrypted with it. A compromise in month twelve exposes the year, not one delivery.
An HSM stores. A proxy transports. A control plane governs.
Sovereign key custody
Exchange context as a first-class object
Granular usage constraints
Immediate multi-party revocation
Evidence with full context
Crypto-agility and native PQC
How it works
The key never moves; the capability does. Every relationship is described with the same five objects, and every invocation is evaluated against all of them before it executes.
The relationship itself: the counterparty, the purpose, the direction of the exchange and the keys in scope. It has a lifecycle of its own (created, active, suspended, closed) and every transition is audited.
The workload that actually invokes: the provider’s integration service, authenticated with the token its own OIDC platform already issues, through workload identity federation. No ANKASecure© credential is handed over.
The function granted (encrypt, decrypt, sign or verify), decoupled from the algorithm and the key that perform it, so you can change both behind the grant.
The key. It stays protected by your HSM or KMS as Root of Trust and is never handed over: the binding grants use, not possession.
The conditions of the grant, enforced at every invocation: a validity window, a total usage cap, a per-minute rate limit and single use.
1 · Authentication
The invoking workload is identified as a known cryptographic actor.
2 · Authorization
It holds a valid grant for this capability, over this key, in this exchange context.
3 · Constraints
The validity window, the usage cap, the rate limit and single use are evaluated.
4 · Mediation
The operation executes in your domain; the key stays under the protection of your Root of Trust.
5 · Evidence
The invocation, its context and its outcome are recorded in a signed, tamper-evident audit trail.
Revocation follows the same path: withdraw a grant, revoke an actor, or suspend or close the exchange context, and the next invocation is refused, without any action by the third party and without re-encrypting data.
Enterprise integration
Exchanging with a third party is two jobs. ANKASecure© changes only the second, and it does not sit in the middle of the exchange: your application protects the information before it is sent, and the third party invokes a governed capability when it needs to use it.
Steps 2 and 3 are the exchange as it works today. Steps 1 and 4 are the only new element: your application calls ANKASecure© to protect the information, and the third party’s application invokes it to decrypt, under your policy. Neither call travels through the channel.
Managed file transfer and SFTP
APIs and integration services
Messaging platforms
Direct application integrations (EDI, connectors)
Counterparty workloads authenticate with the OIDC tokens their own identity platform issues, so no shared secret joins the relationship.
Keys stay protected by your HSM, cloud KMS or PKCS#11 environment. Execution happens in ANKASecure© or in your HSM or KMS, depending on the operation; the Root of Trust stays yours.
Every invocation lands in a signed audit trail that forwards to your SIEM with the signature intact.
Information is not only files: records, JSON or XML payloads, messages and transactions follow the same model. Each third party adapts its integration once to invoke the capability; after that, key rotation and algorithm changes require no further change on either side.
In practice
Four banking scenarios, one per exchange channel. In every one, the existing channel stays in place.
Channel: Batch files
Today
The bank shares a key with each provider, or sends decrypted files. Every provider holds the bank’s cryptographic authority, and a compromise at any one of them exposes the material for all.
With ANKASecure©
One exchange context per provider, each with its own actor and grant. Each batch is protected under its own key, and each provider decrypts only its batch, once, within its window. Providers are isolated from one another, and the bank can cut any of them without coordinating anything.
How it is configured
Crypto-agility
Because the third party invokes a capability rather than holding a key, the key and the algorithm behind it are your lifecycle decisions. A new key every month or a new algorithm every quarter happens in your domain; the third party keeps making the same call and cannot know which key or algorithm fulfils it. Your applications do not change either.
Twelve keys, four algorithms and no line of code changed after the initial integration, on either side. Each delivery is decrypted once, within 48 hours, and a delivery already decrypted cannot be opened again: if the provider is breached, there is no key to steal.
This is a problem of today, not of the quantum era: a key handed to a provider is just as exposed with RSA as with ML-KEM, so it can be solved now, with the algorithms and HSMs already in place. Post-quantum readiness across your external relationships then follows as a consequence: a migration that would be a bilateral project with every counterparty becomes a policy change.
One application in a demo bank, its counterparties, and the sensitivity of every flow between them. Selecting an application names each exchange context — the PSD2 token issuance, the consent revocation — and who is on the other end of it.
Fifteen minutes
Live, on a real deployment, with a counterparty of the kind you have. Nothing in this list is a mock-up or a roadmap item.
An exchange context for a provider: counterparty, type, the keys in scope.
A grant with its constraints: decrypt only, one key set, a two-week window, a usage cap.
The provider decrypts within scope. Then asks for an operation outside it, and is refused with the reason.
The grant is revoked. The very next call from the same workload is refused. Nothing was re-encrypted.
The audit trail: every use, the constraint decision, the revocation — signed, in a chain that would show any edit.
Your provider contract, discussed on the deployment rather than on a slide.
Business value
-70%
Onboarding time
An exchange context is defined, instead of a key exchanged and a pipeline built.
0
Keys shared
With providers, BPOs or external integrations: key exposure is eliminated, not reduced.
< 1 min
Effective revocation
On an incident, without coordinating with the third party and without re-encrypting data.
100%
Auditable evidence
Of every cryptographic operation, with full context, generated in your domain.
Regulatory alignment
Sovereign exchange is not designed around any single regulation, but around the principle they share: the data owner remains responsible for how third parties use its data, and must be able to demonstrate control. Implemented architecturally, alignment with each framework follows.
DORA
PCI-DSS 4.0
SOX
GDPR and national equivalents
SAMA Circular 482021280
CNSA 2.0
Most frameworks ask the owner to demonstrate control, not merely to assert it. Evidence generated by your infrastructure for every cryptographic operation is qualitatively different from a contractual attestation.
Third-party risk frameworks expect you to be able to terminate access when a provider is compromised or a relationship ends. Here it is a structural property, not a procedure that depends on the counterparty.
Supervisors have begun to ask explicitly about third-party dependencies in cryptography. The Saudi Central Bank’s circular on quantum computing risks, in force since August 2026, requires institutions to identify them for their priority cryptographic assets.
What it rests on
Shipping
Exchange contexts and counterparty types
The relationship as an entity with its own lifecycle — created, active, suspended, closed — every transition audited.
Shipping
Capability grants and constraint policies
Operation and key scope, plus four enforced constraints: time window, usage cap, rate limit, single use.
Shipping
Workload identity federation
A counterparty workload authenticates with the token its own OIDC platform issues; no ANKASecure© credential is handed over.
Shipping
Immediate revocation
A revoked grant, a suspended exchange or a deleted application stops authorizing on the next call, with no re-encryption.
Shipping
Signed audit trail
Every use recorded with its grant and constraint decision, tamper-evident, forwardable to your SIEM with the signature preserved.
Shipping
Role model
Platform, tenant and operator authority kept apart, so the person who grants is not the person who operates.
The framework
The Cryptographic Control Plane reference architecture describes each pillar through the scenarios it must handle and one practical test. One of the sovereignty pillar’s scenarios is this solution, word for word.
02
Primary pillar
Scenario realised
Maintaining revocable cryptographic authority over third parties
Practical test
“Can the organization govern policy across all its trust domains without moving keys, and revoke a third party’s cryptographic authority without re-encrypting the data?”
04
Primary pillar
Scenario realised
Enforcing a new enterprise security requirement
Practical test
“Can the organization prove, by querying the control plane, which policy applied to a given operation, and change that policy centrally?”
05
Supporting pillar
Scenario realised
Interoperating with external cryptographic ecosystems
Practical test
“Can the control plane be adopted without replacing the identity, KMS/HSM, PKI, or monitoring platforms already in place?”
Maturity move
A control plane has to be in place first. On top of it, governing counterparties by grant rather than by key is what the reference architecture calls agile operations and full governance: the authority you extend outward is as revocable as the one you keep inside.
It is not a file-transfer product. It does not replace SFTP, your APIs, your messaging or your MFT platform; it governs the cryptographic operations behind them.
It does not remove integration. Your application protects the information before it is sent, and each third party adapts its integration once to invoke the capability instead of decrypting with a key of its own.
It does not protect data you send to the counterparty in the clear, and it does not reach what the counterparty has already decrypted: that remains under the contractual and data-protection obligations of the relationship.
It does not manage the counterparty’s own systems. It decides what they may do with your keys; it does not see what they do afterwards with the result.
It does not certify you. ANKASecure© maps its controls to NIST CSWP 39, the GSA PQC Buyer’s Guide and OWASP and hands you the evidence; the determination is still your auditor’s.
Adoption path
Not a cryptographic inventory and not a post-quantum roadmap: the list of your third parties. Procurement, risk and security teams already know which counterparties receive keys or files today. Adoption goes one exchange context at a time, with the algorithms and HSMs already in place.
1 · 2–4 weeks
An inventory of third-party exchanges, not of the whole cryptographic estate, and the most acute case for the pilot: the highest risk, the most regulatory pressure, or the most frequent need to revoke.
2 · 6–8 weeks
One exchange context in production, keys anchored in your existing HSM, evidence flowing to your audit infrastructure.
3 · 3–6 months
The remaining third parties, onboarded progressively and integrated with your HSM or KMS and identity provider; per-context policies replace per-counterparty key exchanges.
4 · 6–12 months
The control plane governs all the cryptography in your exchanges, with key rotation and post-quantum migration by policy across every external relationship.
Whitepaper
Why control ends at the moment of sharing, the three properties and five building blocks of sovereign exchange, four banking scenarios, regulatory alignment, business value and an adoption path. 38 pages, October 2026.
Read it online →Download the PDF
The download opens on this page as soon as the form is submitted. We ask for a corporate account so we can route follow-up to the right team.
After this
No new deployment, no second contract for the platform. The grant mechanism that governs a counterparty is the one that governs your own teams.
The one where they hold your key. Fifteen minutes on a live deployment to replace the key with a permission you can revoke.
Book the demo