Subprocessors
Last updated: 23 May 2026
Meta-Flux engages the following third parties to deliver the Meta-Flux Platform and operate our business. A data processing agreement (DPA), and, where customer data is in scope, a business associate agreement (BAA), is in place with each. We notify affected customers at least 30 calendar days before a new subprocessor begins processing their data.
Platform infrastructure
The only subprocessor that unconditionally processes customer data on the Meta-Flux Platform.
| Subprocessor | Services used | Region | Compliance |
|---|---|---|---|
| Amazon Web Services | Compute, storage, database, identity, CDN, AI inference (Bedrock) | United States (us-east-1) | SOC 1/2/3, ISO 27001, ISO 27017, ISO 27018, HIPAA-eligible |
AI inference for customer data runs by default through AWS Bedrock. Bedrock-hosted models run within the AWS boundary and inherit the AWS DPA and BAA; there is no transit outside the AWS boundary on this default path.
Optional platform LLM providers (per-tenant opt-in)
These vendors receive customer prompt content only when a customer explicitly configures them as their LLM provider in place of the default AWS Bedrock path. Default state for every tenant is disabled, enforced at the runtime egress boundary.
| Subprocessor | Services used | Region | Compliance |
|---|---|---|---|
| Anthropic | Claude API for LLM inference | United States | SOC 2 Type II, ISO 27001 |
| OpenAI | Chat Completions API for LLM inference | United States | SOC 2 Type II, ISO 27001 |
| Google Gemini | Gemini API for LLM inference (AI Studio) | Global | SOC 2 Type II |
Internal-productivity use of LLM tools. Meta-Flux personnel also use Anthropic Claude (Team plan), OpenAI ChatGPT (Team plan), and Google Gemini (within Google Workspace Business, not the AI Studio API) as internal productivity tools, for drafting, summarisation, and presentation assistance. Each workspace is on a paid Team/Business tier whose default terms exclude customer content from being used to train the vendor’s models, and chat content on these tiers is not retained for training purposes. Any Client Confidential Information referenced in this context is governed by the personnel-access controls described under Authorized Meta-Flux personnel below, not by the per-tenant runtime opt-in described in the table above. No customer-derived experimental data, Results, or PHI is placed into these surfaces.
Business operations and platform research
These vendors do not connect to the Meta-Flux Platform runtime. They are used either in our internal business operations, email, documentation, meetings, messaging, where Client Confidential Information may be incidentally referenced, or as platform-scope research tools where queries are redacted and audit-logged to exclude customer-derived data.
| Subprocessor | Activities | Region | Compliance |
|---|---|---|---|
| Google Workspace | Email, file storage, document collaboration, video conferencing, calendaring | Global | SOC 1/2/3, ISO 27001, HIPAA via BAA |
| Microsoft 365 | Microsoft 365 productivity applications (Word, Excel, PowerPoint) | United States, EU | SOC 1/2, ISO 27001, HIPAA via BAA |
| Notion | Internal documentation and knowledge management | European Union | SOC 2 Type II, ISO 27001 |
| Slack | Internal team messaging | European Union | SOC 2 Type II, ISO 27001 |
| Granola | Meeting transcription for internal meetings | European Union | Per vendor security documentation |
| Brave Search | Web-search API for platform-scope research (no customer data routed) | United States | SOC 2 Type II |
Reference data sources (not subprocessors)
The Meta-Flux research agent queries public scientific databases using only generic biological identifiers (gene symbols, KEGG IDs, PubMed IDs, UniProt accessions). No customer-derived experimental data is included in these queries; these services are not subprocessors under GDPR.
- NCBI / PubMed (US National Institutes of Health)
- OpenAlex (OurResearch)
- Crossref
- OpenTargets (EMBL-EBI)
- Reactome (EMBL-EBI / OICR)
- KEGG (Kyoto Encyclopedia of Genes and Genomes)
- ChEMBL (EMBL-EBI)
- UniProt (UniProt Consortium)
Authorized Meta-Flux personnel
In addition to the subprocessors above, authorized Meta-Flux personnel, including engineers, forward-deployed engineers, and customer-success staff, may access a customer’s data on the customer’s behalf when:
- Delivering services commissioned under the Master Services Agreement, such as bespoke analyses, data ingestion, or integration work.
- Troubleshooting platform behaviour at the customer’s request.
- Performing approved internal operations, security review, incident response, compliance audit, under documented Meta-Flux policies.
Meta-Flux personnel are bound by employee confidentiality agreements and Meta-Flux’s data classification and acceptable-use policies. Every access is recorded in AWS CloudTrail and reviewable on request. Meta-Flux personnel are part of Meta-Flux’s own processing role under the MSA; they are not third-party subprocessors and therefore do not appear in the list above.
Stay informed about changes
We notify affected customers at least 30 calendar days before adding a new subprocessor that will process their data. To receive notifications, ensure your designated security contact is on file with us at onboarding. Objections to a proposed subprocessor must be raised within the notice window by your named security contact.
For questions about our security or compliance posture, or to register an objection, email hello@meta-flux.com.
© 2026 Meta-Flux Limited · meta-flux.com · hello@meta-flux.com
Back to home