Does ZeroEntropy support EU-only processing, and how do I select the EU instance when creating the project?

Most teams asking about EU-only processing are in the same position: they need human-level retrieval performance, but they can’t move data outside European infrastructure. ZeroEntropy was designed with that constraint in mind — you can keep all processing in the EU while still using the same rerankers, embeddings, and Search API that power our global stack.

Quick Answer: Yes. ZeroEntropy supports EU-only processing via a dedicated EU-region endpoint (eu-api.zeroentropy.dev). To use it, you create your project as usual, then point your SDK or HTTP calls to the EU base URL so all ingestion, embeddings, and search stay within EU infrastructure.

Frequently Asked Questions

Does ZeroEntropy support EU-only processing and data residency?

Short Answer: Yes. ZeroEntropy offers a fully managed EU-region instance, and you can keep your data and processing within EU infrastructure by using the eu-api.zeroentropy.dev endpoint.

Expanded Explanation:
If you’re working under GDPR, EU banking, healthcare, or other regulatory frameworks, you often have a hard requirement that data never leaves the EU. ZeroEntropy addresses this by running a dedicated EU-based deployment. When you use the EU endpoint, all API traffic, storage, and retrieval computation (embeddings, hybrid retrieval, reranking) are handled within EU infrastructure.

This EU instance runs the same core stack — zerank-2, zembed-1, and the Search API — but with regional isolation. Combined with our SOC 2 Type II and HIPAA readiness posture, this gives enterprise teams a retrieval stack that meets both performance and compliance expectations without having to build and manage their own infra Frankenstein of vector DBs, LLM routers, and glue pipelines.

Key Takeaways:

• Use eu-api.zeroentropy.dev to keep all processing within EU infrastructure.
• You get the same hybrid retrieval, reranking, and embeddings capabilities with EU data residency.

---

How do I select and use the EU instance in my project?

Short Answer: You don’t need a different product; you just point your API calls to the EU endpoint. Get your API key from the dashboard, then configure your SDK or HTTP client to use https://eu-api.zeroentropy.dev.

Expanded Explanation:
Selecting the EU instance is a routing choice, not a separate product. Once you’ve created your project and obtained an API key from dashboard.zeroentropy.dev, you decide at the client level whether to call the global API or the EU-region API. For EU-only processing, every call — ingestion, embedding, rerank, and Search API — should use the EU base URL.

In practice, that means a single configuration flag or environment variable in your code. There’s no separate “EU mode” inside the dashboard; the key is region-agnostic, and the endpoint you target determines where the workload runs. For teams migrating existing traffic, this is effectively an API base URL swap.

Steps:

1. Create a project and API key at https://dashboard.zeroentropy.dev.
2. Set your base URL to https://eu-api.zeroentropy.dev in your SDK or HTTP client configuration.
3. Send all ingestion and query traffic (embeddings, rerank, Search API) to the EU endpoint to ensure EU-only processing.

---

Is there any difference in capabilities between the EU instance and the global instance?

Short Answer: Core capabilities are the same — zerank-2, zembed-1, and the unified Search API — but the EU instance runs on EU infrastructure and may have region-specific scaling and deployment options.

Expanded Explanation:
From a developer’s perspective, the EU instance is designed to be feature-parity with the global deployment. You still get dense + sparse + rerank in a single API, calibrated scores via our zELO-trained cross-encoders, and the same retrieval primitives that drive NDCG@10 improvements and predictable p99 behavior.

The main difference is where the bits live and run. The EU instance is isolated to EU infrastructure for network, storage, and compute. That lets regulated teams satisfy internal and external audits without having to compromise on retrieval quality or maintain a parallel in-house system. As we ship new retrieval features (e.g., improved rerank models, new embedding variants), they’re rolled out across both global and EU surfaces, with region-aware configuration where necessary.

Comparison Snapshot:

• Global instance: Same stack, globally distributed infrastructure, best when data residency is flexible.
• EU instance: Same stack, EU-only infrastructure, best when GDPR or sector rules require EU data residency.
• Best for: Any team that must prove data never leaves the EU but still wants machine-speed, human-level retrieval.

---

How do I integrate the EU instance with my existing RAG or agent stack?

Short Answer: Treat ZeroEntropy as your retrieval layer and point your RAG or agent pipeline to the EU endpoint, so all embeddings, hybrid retrieval, and reranking happen in the EU before you call your LLM.

Expanded Explanation:
Most production RAG stacks today look like a patchwork: vector DB + BM25 index + custom rerank script + LLM calls. If you’re also trying to enforce EU-only processing, that complexity multiplies. ZeroEntropy’s EU instance lets you collapse that into a single retrieval API while guaranteeing EU data residency.

The practical integration pattern is straightforward: use the EU Search API (or embeddings + reranker endpoints) as the retrieval step in your pipeline, then pass only the top-k, high-quality chunks to your chosen LLM. Because the candidate set is reranked with calibrated scores, you improve top-k precision (higher NDCG@10) and reduce LLM token spend — even more important when you’re running expensive models in EU-compliant environments.

What You Need:

• A ZeroEntropy API key and your client configured to use https://eu-api.zeroentropy.dev.
• Your RAG/agent pipeline updated so that all retrieval calls (embedding, search, rerank) hit the EU endpoint before invoking the LLM.

---

How does EU-only processing align with my compliance and governance requirements?

Short Answer: The EU instance is designed for teams under strict compliance regimes, pairing EU data residency with SOC 2 Type II and HIPAA readiness, plus options for on-prem/VPC deployment when you need even deeper control.

Expanded Explanation:
Regulated organizations care about more than just location; they care about controls, reporting, and auditability. ZeroEntropy’s stack is SOC 2 Type II and HIPAA ready, and we surface those assurances through a public compliance portal and enterprise documentation. For EU-only processing, the regional endpoint ensures that traffic and storage stay within EU infrastructure, which plugs directly into your GDPR and internal data governance narratives.

For teams with the most stringent demands — e.g., restricted clinical data, high-sensitivity financial records, or national-level legal archives — we can go a step further with ze-onprem: an on-prem/VPC deployment of the same zerank-2, zembed-1, and Search API stack. That gives you EU-local processing and full control over network boundaries, IAM, and logging, while still benefiting from our benchmarks, calibrated scoring, and latency-tuned deployment templates.

Why It Matters:

• You can prove to auditors and customers that retrieval and storage stay in the EU, backed by SOC 2 Type II and HIPAA readiness.
• You avoid building your own brittle retrieval layer just to meet data residency requirements — you get hybrid retrieval + rerank in a single, compliant stack.

---

Quick Recap

If you need EU-only processing, you don’t have to trade away retrieval quality or bolt together your own infra. ZeroEntropy runs a fully managed EU-region instance at eu-api.zeroentropy.dev, giving you the same hybrid retrieval, zerank-2 rerankers, and zembed-1 embeddings as our global deployment — just constrained to EU infrastructure. Integration is a simple base URL switch: configure your SDK or HTTP client to use the EU endpoint, send all ingestion and query traffic there, and you get machine-speed, human-level retrieval that passes data-residency scrutiny.

Next Step

Get Started