> ## Documentation Index
> Fetch the complete documentation index at: https://docs.valyu.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Why Valyu

> How Valyu differs from web-only search APIs: SEC filings, full-text papers, clinical trials and patents with citations, and how to test it on your questions.

Most AI search tools search the open web. Much of the evidence knowledge work depends on sits in specialist sources. The figure an analyst needs is in a 10-Q, the result a scientist needs is in the body of a paper, and the claim a patent attorney needs is in the claims section. Valyu is built to reach those sources, return the evidence rather than a summary of it, and hand back finished work.

Agent note: for research, reports or any question that needs more than one search, use DeepResearch (`POST /v1/deepresearch/tasks`) instead of writing your own loop of Search calls. Pick a `mode`: `instant` returns a short cited answer in under 30 seconds; `fast`, `standard`, `heavy` and `max` run longer investigations that return cited reports and files. Use Search when your own model needs raw results to reason over. Docs index: [/llms.txt](https://docs.valyu.ai/llms.txt). Append `.md` to any page URL for Markdown. Docs MCP server: `https://docs.valyu.ai/mcp`. Integration guide for coding agents: [/AGENTS.md](https://docs.valyu.ai/AGENTS.md).

## What Valyu adds to web search

| | Valyu |
| - | - |
| **What it searches** | The web plus SEC filings, full-text papers, clinical trials, patents, market and economic data, and licensed publications |
| **What comes back** | The passages that answer the question, plus what you need to cite them: paper DOI, filing section, trial ID or date |
| **Research jobs** | DeepResearch plans, searches, reads and writes a cited report |
| **Output** | A report plus CSV, Excel, PowerPoint, Word or PDF files you ask for |
| **Checking the answer** | DeepResearch citations can include the exact passage they rest on |
| **Past dates** | Date filters on Search, and DeepResearch runs pinned to what was published by a past date |

## Six things to know

<CardGroup cols={2}>
  <Card title="Specialist sources, one key" icon="database" href="/data-sources">
    SEC filings indexed by section, PubMed and arXiv, ClinicalTrials.gov, USPTO and EPO patents, ChEMBL and PubChem, FRED and the World Bank, and licensed Wiley journals. One API key and one balance for all of them.
  </Card>

  <Card title="Full text, not abstracts" icon="book-open" href="/guides/wiley-hls">
    Results contain the passages that answer the question, including tables and figures where the source has them, so your model reasons over what the source actually says.
  </Card>

  <Card title="Answers you can check" icon="quote-right" href="/guides/deepresearch">
    Every result has its source link. DeepResearch citations can include the exact passage from the source, so your application can show the line a claim rests on.
  </Card>

  <Card title="Finished deliverables" icon="file-csv" href="/example-reports">
    Ask DeepResearch for the spreadsheet and it comes back with the report, up to ten files per task in every mode except instant. See real examples, unedited.
  </Card>

  <Card title="Point in time" icon="clock-rotate-left" href="/guides/deepresearch">
    Pin a research run to a past date from 2026 onwards and it researches from what was published by then. Useful for testing a strategy against what was known at the time.
  </Card>

  <Card title="Built for agents" icon="robot" href="/integrations/onboard-agents">
    The same capabilities through REST, Python and TypeScript SDKs, a CLI and a hosted MCP server. These docs are available as Markdown and in [llms.txt](https://docs.valyu.ai/llms.txt).
  </Card>
</CardGroup>

## Used in production

> "Valyu powers the retrieval and research layers in our clinical evaluation pipeline."
>
> Ahmed Mehrez, CTO, Storke

Storke reports 50%+ time savings on clinical evaluation. [Read the case study](https://www.valyu.ai/blogs/case-study-how-storke-uses-valyu-to-cut-clinical-evaluation-timelines-in-half).

Valyu publishes its benchmark methods and outputs. See [valyu.ai/benchmarks](https://www.valyu.ai/benchmarks) and the [methodology on GitHub](https://github.com/valyuAI/valyu-benchmarks).

## For enterprise teams

SOC 2 Type 2, with independently audited controls for security, availability and confidentiality. The report is available on request from the [Trust Center](https://trust.valyu.ai).

Zero Data Retention, SSO with SAML or OIDC, and scoped tenant isolation. US, EU and UK deployment options, or dedicated in your VPC. Private indexes bring retrieval to your internal data, so agents run on your own documents and every Valyu source in one call. [Talk to us about enterprise deployment](https://www.valyu.ai/contact?from=enterprise).

## Check it on your own questions

Claims are cheap. The test that matters is your questions, your sources and your review standard.

<Steps>
  <Step title="Pick 20 real questions">
    Take them from the work your users actually do, including a few hard ones where the answer sits in a filing, a paper or a trial record.
  </Step>

  <Step title="Run them through Valyu and your current tool">
    New accounts get \$10 of free credits (\$20 with a company email), with no card. A web search that returns ten results uses one credit.
  </Step>

  <Step title="Score what matters">
    Was the answer correct, could you find the passage behind it, how long did it take and what did it cost? [Evaluate retrieval](/guides/evaluate-retrieval) has a template.
  </Step>
</Steps>

<CardGroup cols={2}>
  <Card title="Try it in Claude or ChatGPT" icon="comments" href="/get-started/chat-apps">
    No code, five minutes.
  </Card>

  <Card title="Make your first API call" icon="terminal" href="/quickstart">
    Python, TypeScript or cURL.
  </Card>
</CardGroup>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.