Inherent
Technical docs for Inherent's private company brain. Connect private company context, retrieve it through REST or MCP, and keep every answer grounded with citations and auditability.
The Problem
You're building RAG-powered apps for clients. Every project needs the same pipeline: parse documents, chunk them, embed, index, search. And every project, you rebuild it from scratch.
- Ingestion is a time sink — parsing PDFs, handling edge cases, managing embeddings
- Retrieval is a black box — same query, different answers, no way to debug
- No audit trail — when a client asks "where did the AI get that?", you can't answer
How It Works
Inherent sits between your data sources and your AI. Three layers, one API:
| Layer | What it does | Why it matters |
|---|---|---|
| Truth | Stores every document as the authoritative source (PostgreSQL) | Single source of truth — no stale data |
| Memory | Semantic search across all documents (Weaviate) | Fast, accurate retrieval in milliseconds |
| Audit | Logs every retrieval with full context | Reproduce any result, debug any answer |
Upload a document and search it — that's the whole API:
# Upload a document
curl -X POST https://api.inherent.sh/v1/documents \
-H "X-API-Key: $INHERENT_API_KEY" \
-F "file=@./product-specs.pdf" \
-F "workspace_id=ws_abc123"
# Search across your knowledge base
curl -X POST https://api.inherent.sh/v1/search \
-H "X-API-Key: $INHERENT_API_KEY" \
-H "Content-Type: application/json" \
-d '{"query": "authentication requirements", "workspace_id": "ws_abc123"}'
Key Features
- Managed Ingestion — Upload files via API. Inherent handles parsing, chunking, embedding, and indexing. No pipeline to build.
- Reproducible Retrieval — Same query, same result, every time. No more "it worked yesterday."
- Full Audit Trail — Every retrieval is logged: what was served, when, to whom. Essential for compliance and debugging.
- Multi-Tenant Workspaces — Each client gets a fully isolated workspace. Built for agencies shipping to multiple customers.
Quick Links
| Quickstart | Upload a document and search it in 5 minutes |
| Authentication | API keys, permissions, and rate limits |
| Uploading Documents | Supported file types and async processing |
| Searching | Search parameters and best practices |
| MCP Server | Connect your AI assistant directly to your knowledge base |
| API Reference | Complete endpoint documentation |
Technical Positioning
Use these docs when you need technical proof, implementation details, or self-hosted deployment guidance. If you want to evaluate fit for an internal copilot, agency workflow, or governed company retrieval stack, book a demo.