Sub-processors
Last updated: August 24, 2026 · Version 2026-08-24 · Supersedes the August 23, 2026 version
AKAXA uses the following sub-processors to provide, secure, monitor, and improve the AKAXA platform. Sub-processors process Customer data only as needed to deliver AKAXA Services, under each provider's published data processing terms. AKAXA has not negotiated a bilateral data processing or confidentiality agreement with any sub-processor, and does not represent that it has.
The “AI Training Use” column reflects each provider's published commercial API terms. AKAXA does not present zero data retention (ZDR) as an executed commitment: providers may retain content for a limited period for abuse monitoring under their own terms. Where a provider-specific zero-retention configuration is in place for AKAXA, it will be named here individually.
AKAXA does not operate its own data centre and does not guarantee that data is stored in, or restricted to, any single jurisdiction. A specific data residency position must be agreed in writing for a deployment before confidential material is accepted.
Authorized Sub-processors
| Sub-processor | Purpose | Data Location | AI Training Use |
|---|---|---|---|
| Anthropic | AI analysis, reasoning, and drafting support for AKAXA reports. | Provider-operated regions, primarily United States. | Customer data submitted through AKAXA's API/contracted use is not used to train Anthropic models. |
| OpenAI | AI analysis and drafting through the LLM API, and structured processing for selected AKAXA workflows. | Provider-operated regions, primarily United States. | Customer data submitted through AKAXA's API/contracted use is not used to train OpenAI models. |
| Google (Gemini) | AI analysis or evidence-processing support when enabled for a workflow. | Provider-operated regions. | Customer data submitted through AKAXA's API/contracted use is not used to train Google models. |
| Perplexity | Web research enrichment and source discovery for analysis workflows. | Provider-operated regions, primarily United States. | Customer data submitted through AKAXA's API/contracted use is not used to train Perplexity models. |
| Cloudflare | CDN, WAF, DDoS protection, DNS, and R2 object storage for uploaded files, generated reports, and related artifacts. | Provider-operated regions. | Storage data is not used by AKAXA or Cloudflare for AI model training. |
| Railway | Backend application hosting and production service infrastructure. | Provider-operated regions, primarily United States. | Infrastructure telemetry and hosted data are not used by AKAXA for AI model training. |
| Vercel | Frontend hosting, build, and edge delivery for the AKAXA web application. | Provider-operated regions. | Frontend hosting data is not used by AKAXA for AI model training. |
| Sentry | Error monitoring, stack traces, release health, and operational diagnostics. | Provider-operated regions. | Error telemetry is not used by AKAXA for AI model training. |
| Resend | Transactional email delivery (account, security, and service notifications). | Provider-operated regions. | Email delivery data is not used by AKAXA for AI model training. |
| GitHub | Source control, code review, and CI/CD. The scheduled production-database backup runs pg_dump inside a GitHub-hosted runner before the dump is uploaded to Cloudflare R2, so GitHub infrastructure holds a full copy of the database in transit. | Provider-operated regions. | Repository and CI data are not used by AKAXA for AI model training. |
Changes to This List
AKAXA will provide notice of material sub-processor changes according to the applicable customer agreement or Data Processing Agreement. Customers with contractual notice rights may object within the stated notice period.
Contact
Questions about sub-processors may be sent to privacy@akaxa.io or security@akaxa.io.