Data Intelligence Pipeline
Turn messy, sensitive documents into RAG-ready knowledge graphs. One call. Structured Data Packages. Production-grade quality metrics.
latence-python SDK
v0.2Pipeline-first SDK. Submit files, get structured Data Packages.
Quick Start
Create an account and generate an API key from your dashboard.
Create Account →pip install latencejob = client.pipeline.run(
files=["doc.pdf"]
)pkg = job.wait_for_completion()
print(pkg.document.markdown)
pkg.merge(save_to="out.json")The Pipeline
Core ProductSubmit documents. Get back structured, high-quality data packages ready for RAG, agents, and LLM workflows.
Just provide files. The intelligent default pipeline runs Document Intelligence → Entity Extraction → Knowledge Graph automatically. No configuration required.
Not raw JSON dumps. Organized sections with document markdown, entities, knowledge graphs, quality metrics, and confidence scores.
Services execute as a directed acyclic graph -- independent branches run in parallel. Track per-stage progress with real-time callbacks. Resumable on partial failure.
Download results as an organized ZIP archive with markdown documents, entity JSON, knowledge graph data, quality reports, and a human-readable README.
Merge all outputs into a single, document-centric JSON with zero redundancy. One call to pkg.merge() and you have production-ready data for downstream consumption.
Authentication
All API requests require authentication using a Bearer token.
Rate Limits
API requests are rate-limited per API key, per service. Limits apply uniformly — there are no tier-based rate limits.
| Service | Rate Limit |
|---|---|
| Document Intelligence | 500 req / min |
| Entity Extraction | 1,000 req / min |
| Relation Extraction / Knowledge Graph | 500 req / min |
| Redaction | 1,000 req / min |
| Compression | 1,500 req / min |
| Embed (unified) | 1,500 req / min |
| Embedding (dense) | 2,000 req / min |
| ColBERT | 1,000 req / min |
| ColPali | 1,000 req / min |
| Chunking | 5,000 req / min |
| Dataset Intelligence | 100 req / min |
| Enrichment (Coming Soon) | 2,500 req / min |
Rate Limit Headers
x-ratelimit-limitMaximum requests allowed in the windowx-ratelimit-remainingRequests remaining in the current windowx-credits-usedCredits charged for this requestx-credits-remainingYour remaining credit balanceError Codes
Standard HTTP error codes with additional context.
| Code | Name | Description |
|---|---|---|
| 400 | Bad Request | Invalid request parameters |
| 401 | Unauthorized | Missing or invalid API key |
| 402 | Insufficient Credits | Your credit balance is too low |
| 429 | Too Many Requests | Rate limit exceeded |
| 500 | Internal Server Error | Unexpected server error |
{
"error": "Rate limit exceeded",
"details": "Maximum 500 requests per 60000ms",
"retry_after": 60
}Experimental / Developer APIs
Self-ServiceDirect access to individual services for development, testing, and custom workflows.
These endpoints are available for development and testing. For production workloads, use the Data Intelligence Pipeline above -- it provides structured Data Packages, quality metrics, and is covered by Enterprise SLAs.