This encryption RFP template and data encryption RFP checklist gives you a vendor-neutral set of requirements for protecting data at rest in cloud, SaaS and database platforms with keys that you control. It covers eleven requirement groups (scope, encryption models, key control and sovereignty, platform integrations, availability, cryptography, access policies, audit, regulatory mapping, deployment and exit, vendor and support), a weighted scoring model, 20 vendor questions, red flags, a proof-of-concept plan and contract clauses. Every table can be copied into a spreadsheet.
Table of Contents
- Who this is for and how to use it
- Before you write the RFP
- The requirements checklist
- Weighted scoring model
- 20 vendor questions to ask in the response
- Red flags in vendor answers
- Proof-of-concept plan with pass/fail tests
- Contract and exit clauses
- Copy-ready RFP introduction
- FAQ
- Conclusion
- References
Who this is for and how to use it
This template is for security architects, CISOs, data protection officers, procurement and cloud platform owners who must buy, or re-tender, an enterprise data encryption solution in which the customer, not the cloud or SaaS provider, controls the keys. It applies whether you call the result BYOK, HYOK, external key management (EKM), double key encryption (DKE), client-side encryption, tokenization or field-level encryption.
How to use it:
- Copy the introduction at the end into your RFP and adapt the scope.
- Paste the requirement tables into a spreadsheet. Keep the ID column: it lets you score vendors and trace evidence later.
- Delete requirements that do not apply (for example, the Salesforce rows if you do not use Salesforce) and re-weight the scoring model.
- Ask for evidence, not assertions. Each row names the evidence to request.
Priorities are Must (disqualifying if absent), Should (scored, a gap is acceptable with a plan) and Advanced (differentiating capabilities that a serious product may offer but that you should not make mandatory unless your risk model needs them).
This page is a procurement aid, not legal advice. Check regulatory wording with your counsel.
Before you write the RFP
Most failed encryption projects fail before the RFP, on scope. Settle these five things first.
1. Inventory and classify the data. List the platforms that hold regulated or confidential data: Microsoft 365 (Exchange, SharePoint, OneDrive, Teams), Salesforce, ServiceNow, AWS storage and databases, Azure, SQL Server, Oracle, Snowflake, Databricks, and AI services such as OpenAI. Attach a classification (public, internal, confidential, regulated) to each and note data volumes and residency constraints.
2. Pick the threat you are buying protection against. Different models answer different questions:
| Threat or goal | What it needs |
|---|
| Cloud provider staff or a breach at the provider reading your data | Customer-held keys the provider cannot use without your key service |
| A foreign legal request addressed to the provider | Keys held under your jurisdiction and a provider that cannot decrypt alone |
| Regulator or auditor asking who can decrypt | Key-use audit trail you control |
| Insider or compromised admin account in your own tenant | Access policies on key use, separation of duties, quorum approvals |
| Lost or stolen storage media | Standard encryption at rest is enough |
3. Decide which encryption model per workload. A single model rarely fits every platform. See Hold Your Own Key (HYOK) for the trade-offs between server-side, BYOK and HYOK, and what BYOK really means.
4. Name the stakeholders. Security architecture, cloud platform owners, application owners, DPO or legal, operations/SRE (they will run the key service), procurement and internal audit.
5. Fix your decision criteria and weights before you read any response. The scoring model below gives a starting point.
The requirements checklist
Each group is a table with four columns: ID, Requirement, Priority and Evidence to request. Requirements are written so that a serious product in this category can satisfy and evidence them.
Group 1. Scope and data classification
| ID | Requirement | Priority | Evidence to request |
|---|
| SC-1 | The solution supports the platforms listed in Appendix A (your inventory) and states per platform which encryption model is used | Must | Platform support matrix with documentation links |
| SC-2 | Protection can be applied by data classification (for example, regulated data keyed separately from internal data) | Should | Configuration walkthrough or demo of per-classification keys |
| SC-3 | Data in transit between platform and key service is protected with TLS 1.2 or higher | Must | Architecture diagram, TLS configuration documentation |
| SC-4 | The vendor states which data is not covered (metadata, search indexes, logs, backups, caches) | Must | Written scope statement per platform |
| SC-5 | Coverage of new and existing data is described, including re-encryption of historical data | Should | Migration procedure, throughput figures from a reference deployment |
Group 2. Encryption models
| ID | Requirement | Priority | Evidence to request |
|---|
| EM-1 | Each supported platform is labelled with its model: provider-managed, customer-managed (BYOK), external key (HYOK/EKM) or client-side | Must | Matrix with the model per platform |
| EM-2 | For external-key models, the vendor explains where the cryptographic operation happens (provider side with a wrapped key, or at your key service) | Must | Data-flow diagram of one encrypt and one decrypt |
| EM-3 | Where the platform offers a double-key option, both keys are required to decrypt (for example Microsoft 365 DKE) | Should | Documentation and a test showing decryption fails without either key |
| EM-4 | Client-side or application-level encryption (SDK or proxy) is available for platforms without native external-key support | Should | SDK documentation, supported languages |
| EM-5 | Tokenization or field-level encryption is available for structured data, with format-preserving options where needed | Should | Feature documentation, sample schema |
| EM-6 | Transparent database encryption (TDE) integration with an external key manager is supported for the databases in scope | Must (if databases in scope) | Supported versions list, install guide |
| EM-7 | Where a platform cannot support customer-held keys, the vendor says so explicitly instead of marketing a weaker model under the same name | Must | Written limitations list |
Group 3. Key control and sovereignty
| ID | Requirement | Priority | Evidence to request |
|---|
| KC-1 | Keys are generated and stored in a key manager under the customer's control (own datacenter, dedicated tenancy or a named jurisdiction) | Must | Deployment options, hosting locations |
| KC-2 | The vendor lists which parties (vendor staff, hosting provider, cloud provider) can access key material or invoke key operations, and under which conditions | Must | Written statement, access-control description |
| KC-3 | The customer can revoke platform access to keys unilaterally, and the effect on data is documented (including cached keys) | Must | Revocation test, documentation of cache lifetimes |
| KC-4 | Key material never leaves the key manager in clear text | Must | Architecture description, HSM or equivalent certification |
| KC-5 | The jurisdiction of the vendor and of the key-hosting entity is stated, with its exposure to foreign legal requests | Must | Legal entity chart, hosting contract extract |
| KC-6 | Root keys can be backed by an HSM certified to FIPS 140-3 (or FIPS 140-2 while valid) or an equivalent assurance level | Should | Certificate numbers |
| KC-7 | Key generation, rotation, disable, destroy and import are supported with documented lifecycle states | Must | Lifecycle documentation |
| KC-8 | Key splitting or multi-party control so that no single administrator or machine holds a complete key | Advanced | Architecture paper, quorum configuration demo |
Keep only the rows you need. "Supported" means documented, tested against current platform versions and supported by the vendor, not "possible with custom work".
| ID | Requirement | Priority | Evidence to request |
|---|
| PI-1 | Microsoft 365: double key encryption for labelled content and/or customer key for service encryption | Should | Setup guide, supported workloads, known limitations |
| PI-2 | Salesforce: Shield Platform Encryption with an external key service | Should | Supported key-use modes, setup guide |
| PI-3 | ServiceNow: customer-controlled key integration | Should | Supported editions and setup guide |
| PI-4 | AWS: external key store (XKS) or KMS-integrated key management for S3, EBS, RDS and similar | Should | XKS proxy compatibility statement, tested services list |
| PI-5 | Azure: Key Vault or managed HSM integration or external key capability for storage and databases | Should | Integration document |
| PI-6 | SQL Server and Oracle TDE through an external key manager (EKM / PKCS#11) | Should | Supported versions, performance figures |
| PI-7 | Snowflake: customer-held key as part of Tri-Secret Secure | Should | Setup guide |
| PI-8 | Databricks and other data platforms: customer-managed keys for storage and workspaces | Should | Setup guide |
| PI-9 | AI services (for example OpenAI enterprise key management): customer-held key registration | Should | Documentation, description of where the key service sits in the data path |
| PI-10 | Standard interfaces for custom needs: KMIP, PKCS#11, REST API | Should | Interface documentation |
| PI-11 | An integration does not require the vendor to receive plaintext business data unless explicitly stated | Must | Data-flow diagram |
External key services put your key manager on the critical path of your cloud data. This group is the one most often under-specified.
| ID | Requirement | Priority | Evidence to request |
|---|
| PA-1 | The vendor documents behaviour when the key service is unreachable per platform: fail-closed (data inaccessible) or fail-open/cached | Must | Behaviour table, outage test results |
| PA-2 | Latency added per key operation is stated for each platform, with the platform's own timeout limits | Must | Benchmark, platform documentation reference |
| PA-3 | A highly available deployment (active-active or active-passive across independent nodes or zones) is supported and documented | Must | HA architecture, failover test |
| PA-4 | Stated availability objective (SLA) for the managed service, or sizing guidance for self-hosted | Must | SLA, runbook |
| PA-5 | Capacity: sustained operations per second tested at your expected peak, with headroom | Should | Load-test report |
| PA-6 | Disaster recovery of the key service itself (backup, restore, regional recovery) with stated RPO and RTO | Must | DR plan, last test date |
| PA-7 | Planned maintenance and upgrades do not interrupt key operations | Should | Upgrade procedure |
| PA-8 | A documented break-glass procedure exists, with its own audit trail and approval | Should | Procedure document |
Group 6. Cryptography and post-quantum readiness
| ID | Requirement | Priority | Evidence to request |
|---|
| CR-1 | AES-256 for data encryption keys; RSA (3072 or higher) or elliptic-curve options for key wrapping where the platform requires them | Must | Algorithm list |
| CR-2 | Algorithms and parameters follow current national guidance (NIST SP 800-57 and equivalents, BSI TR-02102) | Should | Conformance statement |
| CR-3 | Key rotation without service interruption, with automatic and on-demand modes | Must | Rotation demo |
| CR-4 | Post-quantum roadmap: support for ML-KEM (FIPS 203), ML-DSA (FIPS 204) and SLH-DSA (FIPS 205), with dates | Should | Roadmap, current status per algorithm |
| CR-5 | Crypto-agility: algorithms can be changed by configuration, not by re-architecture | Should | Design description |
| CR-6 | Random number generation from a validated source | Must | Documentation, certificate |
| CR-7 | Hybrid (classical plus post-quantum) key establishment option where the platform allows it | Advanced | Product documentation |
Group 7. Access policies
| ID | Requirement | Priority | Evidence to request |
|---|
| AP-1 | Role-based access control for key administration, separated from key use | Must | Role model |
| AP-2 | Policies per key define which application, identity, tenant or service may use it | Must | Policy examples |
| AP-3 | Strong authentication for administrators (MFA, SSO integration with your identity provider) | Must | Authentication options |
| AP-4 | Conditional policies on key use (time, network, identity attributes, rate) | Should | Policy language reference |
| AP-5 | Separation of duties: the person who creates a key cannot alone authorize its destruction | Should | Workflow demo |
| AP-6 | Approval quorum (M-of-N) for sensitive administrative actions | Advanced | Quorum configuration |
| AP-7 | Policy changes are versioned and reviewable | Should | Change history sample |
Group 8. Audit and monitoring
| ID | Requirement | Priority | Evidence to request |
|---|
| AU-1 | Every key operation (use, administration, policy change, failed attempt) is logged with identity, time, key and outcome | Must | Log schema and sample |
| AU-2 | Logs are exportable to your SIEM through standard formats or APIs, in near real time | Must | Integration documentation |
| AU-3 | Logs are protected against alteration; retention is configurable | Must | Description of protection |
| AU-4 | Alerts on anomalous use (volume spikes, unusual source, repeated denials) | Should | Alert rules |
| AU-5 | Behaviour if the audit record cannot be written is defined (refuse the operation vs continue) and configurable by the customer | Should | Documentation of the audit failure mode |
| AU-6 | Tamper-evident audit (hash chaining or signed records) that an auditor can verify independently | Advanced | Verification procedure |
| AU-7 | Reports for auditors that map key controls to your frameworks | Should | Sample report |
Group 9. Regulatory mapping
Ask the vendor to map capabilities to the rules that apply to you, in their own words. Do not accept a logo list. The references below are generic anchors, not a compliance claim.
| ID | Requirement | Priority | Evidence to request |
|---|
| RG-1 | GDPR: support for technical measures under Article 32 (encryption, resilience) and for controller-processor responsibilities under Article 28; documented handling of third-country transfers | Must (EU/EEA data) | Data processing agreement, transfer analysis |
| RG-2 | Swiss FADP: support for data security obligations and cross-border disclosure rules | Must (Swiss data) | Mapping document |
| RG-3 | DORA: support for ICT risk management, including cryptographic controls and key management (Article 9), and ICT third-party risk requirements | Must (EU financial entities) | Mapping document, contractual provisions |
| RG-4 | NIS2: support for the policies on cryptography and encryption required as a cybersecurity risk-management measure (Article 21) | Must (in-scope entities) | Mapping document |
| RG-5 | FINMA: support for outsourcing, operational resilience and data-protection expectations for supervised institutions (Circular 2023/1) | Must (supervised Swiss entities) | Mapping document |
| RG-6 | HIPAA: support for the Security Rule technical safeguards, including encryption of ePHI, and willingness to sign a business associate agreement where applicable | Must (US health data) | BAA template, mapping |
| RG-7 | Independent assurance of the vendor (ISO/IEC 27001 certificate, SOC 2 report or equivalent) with scope covering the key service | Must | Certificate and scope statement |
| RG-8 | Vendor supports regulator and customer audit and inspection rights | Should | Contract clause |
Group 10. Deployment and exit/portability
| ID | Requirement | Priority | Evidence to request |
|---|
| DX-1 | Deployment options: SaaS, dedicated tenancy, self-hosted/on-premises, or hybrid, with the responsibilities of each | Must | Deployment guide |
| DX-2 | Time to deploy for a first platform, with a realistic plan | Should | Project plan from a reference customer |
| DX-3 | Infrastructure-as-code or API automation for key and policy configuration | Should | API/Terraform documentation |
| DX-4 | Exit: keys can be exported or re-wrapped to another key manager in a documented procedure, or data can be re-encrypted under new keys | Must | Exit procedure |
| DX-5 | Exit does not leave data unreadable: the vendor describes the steps and time needed per platform | Must | Exit plan |
| DX-6 | Open interfaces (KMIP, PKCS#11, documented REST API) reduce lock-in | Should | Interface list |
| DX-7 | Key and policy data can be exported in a documented format | Should | Export sample |
Group 11. Vendor and support
| ID | Requirement | Priority | Evidence to request |
|---|
| VS-1 | Support hours and response times by severity, in your time zones and languages | Must | Support policy |
| VS-2 | Named technical contact or customer success for the project | Should | Contract |
| VS-3 | Vulnerability disclosure and patch policy with timelines | Must | Policy |
| VS-4 | Product roadmap relevant to your platforms, shared under NDA if needed | Should | Roadmap |
| VS-5 | Reference customers in your sector and size, contactable | Should | Reference list |
| VS-6 | Financial and operational stability, subcontractor list and change notification | Should | Corporate information |
| VS-7 | Training and documentation for administrators | Should | Documentation set |
Weighted scoring model
Weights should reflect your risk model. The table below is an example with made-up generic numbers.
| Category | Example weight |
|---|
| Key control and sovereignty (Group 3) | 20% |
| Encryption models and platform integrations (Groups 2 and 4) | 20% |
| Performance and availability (Group 5) | 15% |
| Audit and access policies (Groups 7 and 8) | 10% |
| Cryptography and PQC readiness (Group 6) | 5% |
| Regulatory mapping (Group 9) | 10% |
| Deployment and exit (Group 10) | 10% |
| Vendor, support and commercial (Group 11 and price) | 10% |
| Total | 100% |
0 to 5 scale for each requirement:
| Score | Meaning |
|---|
| 0 | Not offered or no answer |
| 1 | Planned but not available |
| 2 | Available with significant custom work or gaps |
| 3 | Available and documented, partial coverage |
| 4 | Available, documented, evidenced in a demo or PoC |
| 5 | Fully meets and exceeds, evidenced by reference customer |
Apply a pass/fail gate first: any Must scored 0 or 1 disqualifies, or requires a written remediation before shortlist.
Worked example (illustrative numbers only). Vendor A averages 4.0 on key control, 3.5 on integrations, 3.0 on availability, 4.0 on audit, 3.0 on cryptography, 4.0 on regulatory, 3.0 on deployment and exit, 3.5 on vendor. Weighted: 0.20 x 4.0 + 0.20 x 3.5 + 0.15 x 3.0 + 0.10 x 4.0 + 0.05 x 3.0 + 0.10 x 4.0 + 0.10 x 3.0 + 0.10 x 3.5 = 0.80 + 0.70 + 0.45 + 0.40 + 0.15 + 0.40 + 0.30 + 0.35 = 3.55 out of 5. Compare it with other vendors scored the same way, and with the price over five years.
20 vendor questions to ask in the response
- For each platform in our scope, which encryption model do you actually deliver (provider-managed, BYOK, HYOK/EKM, client-side)?
- Where does the cryptographic operation take place on a decrypt: at your service, or at the cloud provider with a key you gave it?
- Who, in your organisation and in your hosting chain, can invoke or access key operations? Under what controls?
- In which legal jurisdiction is the key service hosted and operated? Which authorities could compel you or your hosting provider?
- What happens to our data when the key service is unreachable, per platform? Is it fail-closed, and for how long do caches keep working?
- What latency does a key operation add, and what are the timeouts of each platform you integrate with?
- Show us your highest-availability reference architecture and your last failover test.
- What are your RPO and RTO for the key service itself?
- How do we revoke the platform's access, and what is the measured time until data becomes inaccessible?
- How are keys rotated, and does rotation require re-encryption of data?
- Which HSMs or equivalents back root keys, and with which certifications?
- How is administrator access authenticated, and can we require approvals from several people?
- What is logged, where can it be exported, and how do we verify logs were not altered?
- What happens if an audit record cannot be written?
- Which of our regulations (GDPR, FADP, DORA, NIS2, FINMA, HIPAA) have you been assessed against by customers or auditors, and what do you provide for audits?
- What is your post-quantum roadmap, with dates, for key wrapping and signatures?
- Which parts of the solution are your own code, and which are third-party or open-source components?
- Describe the exit: how do we leave in 90 days with all data readable?
- What are the known limitations for each platform integration (features that stop working under customer-held keys)?
- Which customers in our sector run this in production, and how long?
Red flags in vendor answers
- "Encrypted at rest" offered as the whole answer. That is provider-managed encryption; it does not address key control.
- No written statement of who can access keys, or an answer that depends on "trust".
- BYOK sold as HYOK. If the provider ends up holding a usable copy of the key, it is not HYOK.
- Fail-open behaviour presented as a feature, with no explanation of what that does to the control you are buying.
- No latency or capacity figures for the platform that matters most to you.
- High availability described as "cloud-native" without a failover test.
- Exit plan answered with "contact support".
- Compliance claims as a list of regulations with no mapping to capabilities.
- Post-quantum "ready" without named algorithms or dates.
- References that cannot be called.
Proof-of-concept plan with pass/fail tests
Run the PoC on the two or three platforms that carry the most risk. Define pass/fail before the PoC starts.
| Test | Pass criterion |
|---|
| Deploy the key service and connect platform 1 | Encrypt and decrypt succeed within the agreed time; documented steps are reproducible by your team |
| Connect platform 2 (for example a database with TDE) | Same as above, with measured performance overhead within your threshold |
| Revoke access | Data becomes inaccessible within the documented time; restoring access recovers it |
| Key service outage | Behaviour matches the documented fail-closed or cached behaviour; recovery requires no manual re-encryption |
| Failover | Planned node failure without failed operations beyond the stated threshold |
| Key rotation | Rotation completes without service interruption |
| Access policy | An unauthorised identity and an unauthorised application are denied and logged |
| Audit | Operations appear in your SIEM within the agreed delay; an altered record is detected (if tamper-evidence is required) |
| Load | Sustained peak throughput with agreed latency and no errors |
| Exit | Data re-encrypted or re-wrapped under a different key manager in the PoC, with no data loss |
Contract and exit clauses
- Data and key ownership. Keys, key metadata and audit logs belong to the customer.
- Jurisdiction and access. Hosting location, entities with potential access, and an obligation to notify the customer of any legal request unless prohibited by law.
- Subprocessors. List, change notification and objection right.
- Availability. SLA with credits, defined fail-closed or fail-open behaviour and maintenance windows.
- Security. Vulnerability disclosure timelines, breach notification period, audit and inspection rights.
- Regulatory. Clauses addressing the specific frameworks that apply to you (for example DORA contractual provisions for ICT third-party services).
- Exit. Assistance, export formats, transition period, price for exit services, and deletion certificates.
- Escrow or continuity where the key service is critical.
- Termination for convenience or for regulatory change.
Copy-ready RFP introduction
[Company] is seeking proposals for an enterprise data encryption solution that protects data at rest in [list of platforms: for example Microsoft 365, Salesforce, ServiceNow, AWS, Azure, SQL Server, Oracle, Snowflake, Databricks and AI services] using cryptographic keys that remain under [Company]'s control. The solution must allow [Company] to determine who can use and revoke keys, document the behaviour of each platform when the key service is unavailable, provide audit evidence suitable for [applicable regulations: GDPR, Swiss FADP, DORA, NIS2, FINMA, HIPAA] and offer a documented exit path. Respondents must complete the requirement tables in Appendix B, state their assumptions for each platform, and provide evidence for each Must requirement. Respondents will be evaluated with the weighted model in Appendix C and shortlisted vendors will take part in a proof of concept with pass/fail criteria.
FAQ
What should an encryption RFP include?
It should include scope and classification, the encryption models by platform, key control and jurisdiction, integrations, failure behaviour, cryptography, access policies, audit, regulatory mapping, deployment and exit, and vendor support. Add a scoring model and a PoC. The checklist above provides all of these as copyable tables.
What is the difference between BYOK and HYOK in an RFP?
With BYOK you generate or import a key that the provider then holds and uses. With HYOK or external key management the key stays in your key manager and the provider calls it for each operation. Require the vendor to state which model applies per platform, and see BYOK explained.
Should we require fail-closed behaviour?
For regulated data, fail-closed is usually what you want: if the key service is unreachable or refuses, data stays inaccessible. The price is that your key service becomes critical infrastructure. Ask for high availability, outage tests and documented cache behaviour for each platform, and decide fail-closed versus fail-open per workload.
Do we need client-side or double key encryption?
Not for every workload. Double key encryption, such as Microsoft 365 DKE, is meant for a small set of highly sensitive content because some features, such as server-side search, are limited on protected content. Use external key management for the broad estate and double-key or client-side protection for the crown jewels. See the DKE setup guide.
How do we compare HSM-based and software or multi-party key services?
Compare them on the same requirements: who can access keys, certifications, availability, performance, audit and exit. Do not exclude a category by rule unless your regulation does. The KMS RFP template covers key management products in more depth.
How should we handle post-quantum cryptography in the RFP?
Treat it as a roadmap and crypto-agility requirement: ask which algorithms from FIPS 203, 204 and 205 are supported, when, and how algorithms are changed. See the post-quantum RFP template for a dedicated checklist.
Is a vendor outside the EU or Switzerland acceptable?
It depends on your risk assessment, the data and your regulator. What matters is who can access keys and under which legal system. Ask for the jurisdiction of the vendor and of the hosting entity, and read data sovereignty and encryption keys.
Conclusion
A good encryption RFP is a test of control, not of features. Ask who holds the keys at runtime, what happens when the key service goes down, how you revoke, how you prove it to an auditor and how you leave. Score vendors with the same weights, and validate in a PoC. DuoKey provides customer-controlled key management across the platforms above, and you can compare it against others using the comparison pages, Fortanix and Entrust. Include DuoKey, and every other vendor, only on the strength of evidence.
References
- AWS KMS developer guide, External key stores: https://docs.aws.amazon.com/kms/latest/developerguide/keystore-external.html
- Microsoft Learn, Double Key Encryption: https://learn.microsoft.com/en-us/purview/double-key-encryption
- Regulation (EU) 2016/679 (GDPR): https://eur-lex.europa.eu/eli/reg/2016/679/oj
- Regulation (EU) 2022/2554 (DORA): https://eur-lex.europa.eu/eli/reg/2022/2554/oj
- Directive (EU) 2022/2555 (NIS2): https://eur-lex.europa.eu/eli/dir/2022/2555/oj
- Swiss Federal Act on Data Protection (FADP): https://www.fedlex.admin.ch/eli/cc/2022/491/en
- FINMA Circular 2023/1 Operational risks and resilience: https://www.finma.ch
- HIPAA Security Rule, 45 CFR 164.312: https://www.ecfr.gov/current/title-45/subtitle-A/subchapter-C/part-164/subpart-C
- NIST SP 800-57 Part 1 Rev. 5, Recommendation for Key Management: https://csrc.nist.gov/pubs/sp/800/57/pt1/r5/final
- NIST FIPS 203, 204, 205: https://csrc.nist.gov/projects/post-quantum-cryptography
- DuoKey documentation, OpenAI EKM: https://docs.duokey.com/docs/openai-ekm/overview
- DuoKey documentation, Salesforce EKM: https://docs.duokey.com/docs/salesforce-byok/ekm-module
- DuoKey documentation, SQL EKM: https://docs.duokey.com/docs/sql-ekm/overview
- DuoKey documentation, Cockpit security and audit: https://docs.duokey.com/docs/cockpit/security