# What you can build

Five concrete products, each with the endpoints it needs and a link to the guide that builds it.

| Build this | Endpoints | Guide |
|---|---|---|
| A research assistant that answers with citations | Searches, Chats | literature-review-agent |
| "Chat with your library" inside your app | Files, Documents, Chats | grounded-answers |
| A systematic-review data pipeline | Extractions, Exports | batch-extraction |
| A citation/bibliography feature | Papers, Citations | citations |
| A research tool for an LLM agent | tool schemas | tool-use |

## What this API is not for

Knowing the edges saves you a prototype:

- **Not a general-purpose LLM.** Prompts unrelated to the literature or to your uploaded documents
  are refused. There is no free-form completion endpoint.
- **Not a plagiarism checker.** ai-detections estimates whether text reads as machine-generated.
  It does not compare your text against a corpus for copied passages, and its score is not evidence
  of misconduct.
- **Not a full-text mirror.** You may query, quote, and cite. You may not rehost article full text
  or rebuild the corpus from paginated calls.
- **Not a source of clinical, legal, or financial advice.** Answers summarise published literature.
  They are not reviewed by a professional and must not be presented to end users as guidance.
- **Not a real-time index.** New works appear within days of publication, not minutes. If you need
  publication-day coverage, poll the publisher.

> [!danger] Redistribution limits apply
> What you may store, cache, and show to end users depends on each paper's licence. Read
> content-licensing before designing a product around full text.