How do I sign up for ZeroEntropy and create an API key to start testing?

Most teams find the real bottleneck in RAG and AI search isn’t the LLM—it’s retrieval. The fastest way to see if ZeroEntropy fixes that for your stack is to sign up, generate an API key, and send your first rerank or search request in a few lines of code.

Quick Answer: To start testing ZeroEntropy, create an account at the dashboard, verify your email, generate an API key from the console, and then plug that key into the ZeroEntropy SDK or HTTPS calls to hit the rerank, embeddings, or Search API endpoints.

Frequently Asked Questions

How do I sign up for ZeroEntropy and get my first API key?

Short Answer: Go to the ZeroEntropy dashboard, create an account, and generate an API key from the API Keys section to start testing immediately.

Expanded Explanation:
You onboard through the ZeroEntropy dashboard (no sales call required). Once you’ve created and verified your account, you can generate one or more API keys scoped to your projects. Those keys give you instant access to our retrieval stack—zerank-2 for reranking, zembed-1 for embeddings, and the unified Search API.

From there, you can use the official SDK or direct HTTPS calls to start evaluating retrieval quality (NDCG@10, latency, token usage) on your own data. The entire path from sign-up to first successful request typically takes just a few minutes.

Key Takeaways:

• Sign up in the dashboard, then create an API key under API Keys / Settings.
• The same key works across rerank, embeddings, and Search API endpoints.

---

What are the exact steps to start testing ZeroEntropy in my environment?

Short Answer: Create an account, generate an API key, install the SDK, and call the rerank, embeddings, or Search endpoint with a simple test payload.

Expanded Explanation:
ZeroEntropy is built for fast evaluation loops. After you sign up and grab an API key, you can run a smoke test with one query and a few candidate documents, then scale to your own dataset. The typical flow is:

• Validate connectivity and auth with a minimal rerank call.
• Integrate into your existing RAG or search pipeline as an “API swap” where you previously called a generic reranker or vector store.
• Start tracking quality and latency metrics (e.g., NDCG@10, p50/p99) for your queries.

Steps:

1. Sign up on the dashboard

   • Visit https://dashboard.zeroentropy.dev.
   • Create an account with your work email and set a password (or use SSO if enabled).
   • Verify your email address if prompted.

2. Create an API key

   • Log into the dashboard.
   • Navigate to API Keys, Settings, or similar credentials section.
   • Click Create API Key, give it a name (e.g., dev-rag-eval), and copy it securely.
   • Store it in your secret manager or environment variables (e.g., ZEROENTROPY_API_KEY).

3. Run your first request

   • Install the SDK in your language of choice (example in Python):

     pip install zeroentropy
     

   • Use your key to call the API:

     from zeroentropy import ZeroEntropy
     
     zclient = ZeroEntropy(api_key="<YOUR_API_KEY>")
     
     response = zclient.rerank(
         query="Find the indemnification clause",
         documents=[
             "This agreement is governed by the laws of…",
             "The indemnification clause states that…",
             "Payment terms are net 30 days…",
         ],
         model="zerank-2",
     )
     
     print(response)
     

   • Confirm you get ranked results and calibrated scores back, then plug into your RAG or search stack.

---

What’s the difference between using a basic rerank test and the full Search API when I start?

Short Answer: A rerank test checks how well ZeroEntropy orders your existing candidates, while the Search API gives you end-to-end hybrid retrieval (dense + sparse + rerank) from day one.

Expanded Explanation:
If you already have a vector DB or search backend, the fastest way to see value is to call zerank-2 as a standalone reranker on your current candidate set (e.g., top-100 BM25 or vector hits). This isolates the ranking problem and lets you compare NDCG@10 and p99 latency against your existing reranker (Cohere rerank-3.5, Jina rerank-m0, etc.).

If you’d rather not maintain a Frankenstein of BM25 weights, vector thresholds, and custom pipelines, you can start directly with the Search API. It handles dense + sparse retrieval plus reranking in a single endpoint—no separate vector DB, retriever, and reranker orchestration.

Comparison Snapshot:

• Option A: Rerank API only

  • Drop-in replacement for your current reranker.
  • Use your own candidate retrieval (BM25, vector DB, etc.).

• Option B: Full Search API (hybrid retrieval)

  • ZeroEntropy handles dense + sparse retrieval and reranking.
  • Ingestion tokens + OCR included for document-heavy workloads.

• Best for:

  • Rerank-only if you already have infra and want a quick, controlled benchmark.
  • Search API if you want unified retrieval and to stop tuning low-level search knobs.

---

How do I move from “hello world” to a production-like test with my own data?

Short Answer: Ingest a slice of your corpus, wire the API into your RAG/search pipeline, and measure retrieval quality and latency across real queries.

Expanded Explanation:
The first rerank call tells you the system works; the real signal comes from running ZeroEntropy against your actual legal documents, medical guidelines, support tickets, or audit logs. You can start with a narrow slice (e.g., 10k–100k documents), ingest via the Search API, and then run your real query logs through to measure NDCG@10, p50/p90/p99, and LLM token savings.

Because ZeroEntropy is SOC 2 Type II and HIPAA-ready, and offers EU-based managed instances plus on-prem/VPC deployment (ze-onprem), you can also align your evaluation with your compliance requirements from day one.

What You Need:

• A representative sample of your corpus (e.g., PDFs, HTML, docs, ticket logs).
• A small harness or script that:
  • Sends your queries to ZeroEntropy (rerank or Search API).
  • Records metrics (relevance labels or click data, latency, top-k precision).

---

How does starting with the free trial and an API key translate into real business impact?

Short Answer: The trial lets you prove that better retrieval—higher NDCG@10 and stable p99 latency—reduces hallucinations, speeds up workflows, and cuts LLM token spend before you commit.

Expanded Explanation:
ZeroEntropy’s starter plan gives you a free two-week window with 1,000 queries and 1M ingestion tokens to run a realistic evaluation. That’s enough to test against your key workflows:

• Legal clause and precedent retrieval.
• Clinical evidence and guideline lookup.
• Compliance/audit investigations.
• Customer support ticket or knowledge base search.

By plugging the API key into your existing RAG or agent stack, you can compare “before vs. after” on:

• Retrieval quality: NDCG@10 lift vs. your current search or reranker.
• Latency: p50–p99 behavior under your own traffic patterns.
• Cost: Fewer low-quality chunks sent to the LLM → fewer tokens and lower spend.

When you see that the right evidence is now consistently surfaced in the top-k instead of buried at rank 67, you can justify rolling ZeroEntropy into production—either via the hosted API (including EU-region options) or ze-onprem for maximum control.

Why It Matters:

• Better retrieval directly improves answer accuracy and trust in your AI workflows.
• Reranking and hybrid retrieval reduce total RAG costs by sending fewer, higher-quality tokens to expensive LLMs.

---

Quick Recap

To start testing ZeroEntropy, you sign up on the dashboard, generate an API key, and run your first rerank or Search API call in minutes. Begin with a simple “hello world” test, then move quickly to your own corpus and queries to measure NDCG@10, latency, and token savings. Whether you’re swapping out an existing reranker or unifying dense + sparse + rerank via the Search API, the same API key unlocks the entire retrieval stack—with SOC 2 Type II and HIPAA alignment, EU-region options, and on-prem/VPC deployment when you’re ready to go beyond the trial.

Next Step

Get Started