Knowledge API
Ingest your docs. Ask questions. Get a grounded, cited answer — no vector database to run.

How we compare

"Managed RAG" usually means either a vector database you still operate, or an enterprise search platform you adopt. Brainiall ships the primitive: ingest and query, two calls, with retrieval and a grounded cited answer bundled into the query call — billed per call, with a self-serve key.

Prices are list-price approximations for orientation, not quotes. Always check each vendor's current pricing page.

Pricing

Pay for what you do: a small per-1,000-characters charge to ingest, a flat per-query charge to ask. Storage is included. Free tier is generous enough to wire up an end-to-end RAG flow.

PAYG: $0.001 / 1,000 characters ingested · $0.005 / query (retrieval; +nothing extra for the synthesized answer or reranking). Storage included. No minimums, no contracts.

Two calls: ingest, query

# 1. Ingest a document into a knowledge base (your namespace)
POST https://api.brainiall.com/v1/knowledge/my-kb/ingest
  {"text": "<your document text>", "title": "Refund policy"}
  -> {"doc_id": "doc_5a3eda5bc7344f90", "n_chunks": 14, "chars_in": 18230, ...}

# 2. Query it — passages + a grounded, cited answer
POST https://api.brainiall.com/v1/knowledge/my-kb/query
  {"question": "How long do customers have to request a refund?", "top_k": 6}
  -> {"answer": "Customers may request a refund within 30 days of purchase.",
      "found": true,
      "passages": [{"text": "...within 30 days...", "score": 0.81, "doc_id": "doc_5a3...", "cited": true}, ...],
      "synthesized": true}

# Query, retrieval only (no synthesized answer)
POST https://api.brainiall.com/v1/knowledge/my-kb/query
  {"question": "...", "synthesize": false}

# Re-order the retrieved passages by relevance before answering
POST https://api.brainiall.com/v1/knowledge/my-kb/query
  {"question": "...", "rerank": true}

# What's in this knowledge base?
GET https://api.brainiall.com/v1/knowledge/my-kb/documents

Answers are grounded: drawn only from the retrieved passages, with the passages they cite flagged (cited: true). If the passages don't contain the answer, the response says so (found: false) rather than guessing. Namespaces isolate knowledge bases — one per tenant, per project, per document set, whatever fits your model.

What it's for

• Chat over your docs: a help center, a product manual, a knowledge base — ingest once, then answer questions about it with citations users can verify.
• Support deflection: surface the exact policy paragraph that answers a ticket, before it reaches a human.
• Agent grounding: give an agent a retrieval tool that returns passages plus a short grounded answer — no hallucinated facts, every claim traceable to a source.
• Internal search with answers: ingest wikis, runbooks, contracts; query in natural language; get the relevant passages back.
• No infrastructure to run: no vector database to operate, no embedding model to host, no chunking or retrieval code to maintain. Two REST calls.

Press kit & resources

What reviewers, integrators and procurement teams typically ask for.

More specialty APIs

Same single API key, same usage-based pricing, different problem solved.

Agent MemoryThe low-level memory primitive (add / search / recent)

Text IntelligenceSummarization + grounded Q&A over text you supply

Document AIExtract structured fields from scans & PDFs (then ingest them)

Translation100-language neural machine translation