Skip to content
Rivul AI
Explore
FeaturesDeep ResearchRivul MCP
PricingDocsBlogAbout
Sign in
FeaturesPricingRivul MCPDocsBlogAbout
Rivul AI

A free Multimodal AI Research Workspace for documents, recordings, papers and your own library.

Explore

FeaturesDeep ResearchRivul MCP

Resources

DocumentationBlogCompare research tools
About Rivul AIContact and support
© 2026 Rivul AI
PrivacyTerms

Start here

  • QuickstartGet started
  • Bring Your Own KeyGet started
  • What Is MCP for AI Research?Agents
  • TroubleshootingTroubleshooting
  • Limits and QuotasReference

↑↓ to move↵ to openesc to close
Skip to article
Documentation
Overview
  • What is Rivul AI
  • Quickstart
  • Bring your own key
  • Workflows
  • On a phone
  • Papers
    • Citations
    • Claim Checking
    • Analyze
    • Review
    • Export
  • Research
  • Ask Rivul
  • Agents and Runs
  • Skills
  • Rivul MCP
    • Claude Code
    • OpenAI Codex
    • Cursor
    • Google Antigravity
    • Meta Muse Code
    • MCP tool reference
    • REST API reference
    • Keys, apps and activity
    • Usage and fair use
  • Library
  • Search Papers
  • Connectors
  • Document OCR
  • Transcribe
  • Listen
  • Citation Tools
  • Settings
  • Privacy and AI use
  • Export and delete data
  • Common problems
  • Limits and quotas
  • Keyboard Shortcuts
  • Glossary
  1. Documentation
  2. Get started
  3. What is Rivul AI

What Is Rivul AI

Rivul AI is a free multimodal AI research workspace. Bring in PDFs, scans, recordings, text or papers, then ask, research and run agents across them.

Last updated Sep 25, 2026

On this page
  • Who it is for
  • From any source
  • Research across it
  • Write, cite and export
  • Free, with your own AI key
  • Where your data goes
Rivul AI agent run report for What speedups do speculative decoding methods report for LLM inference in 2023-2025, and what limits them?, done and written from 8 sources by the LLM inference scout agent, with its cited report.
A finished agent run: the question on speculative decoding speedups, the eight sources the agent read, and the start of its cited report.

Who it is for

Anyone who works from sources:

Students turn a recorded lecture into a timestamped transcript and ask questions across the week's readings.

Engineers search papers and read their metadata from Claude Code, Codex or Cursor through Rivul MCP, or call the same search over the REST API.

Academics search 300M+ academic papers, keep the ones they use in a Library, and draft with numbered references.

Researchers run agents that search, read and write a cited report, and keep every run and source for review.

Anyone else can get plain answers from a long PDF such as a contract or a manual, listen to an article, or transcribe a voice note or a meeting.

From any source

PDFs and papers: upload a PDF up to 10 MB to the Library, add one by DOI or arXiv ID, import a .bib file, or pull files from Google Drive and Zotero. Scanned PDFs are read with OCR when you add them.

Search papers: search 300M+ academic papers in the Rivul Index, with no AI key.

Scans and images: Document OCR returns the text of PDFs, scans and images, as plain text or Markdown.

Recordings: Transcribe turns audio and video, or a recording made in the browser, into timestamped text.

Text: Listen reads a library paper, a draft or pasted text aloud.

Research across it

Ask Rivul answers questions across your own PDFs and cites each passage it used, with its page.

Research runs a question across the Rivul Index and your Library and writes a cited brief with a findings table.

Agents run a saved research brief on a question: they search, read and write a cited report, and Runs keeps every step. Skills are reusable instructions that guide how an agent works.

Rivul MCP gives your own AI client, such as Claude Code, Codex or Cursor, read-only paper search through one connection.

  • Ask Rivul→
  • Research→
  • Agents and Runs→
  • Rivul MCP→

Write, cite and export

Papers is the editor for a draft, with references from your Library, claim checks against your sources, version history, and export to Word, LaTeX, Markdown, plain text and BibTeX. Citation tools format a single reference from a DOI or a saved paper.

  • Papers
  • Citation tools

Free, with your own AI key

Once you are signed in, every Rivul AI capability is free: search, sources, Ask Rivul, Research, agents, Rivul MCP, citations, and export. You can store up to 5 PDFs at a time; Library shows how many are stored.

AI features run on your own API key from Google Gemini, OpenAI, Anthropic, xAI, Mistral, or Meta, and your provider bills you directly. Search, the Library, Document OCR, Transcribe, Listen and Rivul MCP need no AI key.

  • Bring your own key
  • Limits and quotas

Where your data goes

Your text goes to the AI provider you connected only when you use an AI feature, and only the passages that feature needs. Document OCR, Transcribe, library search and the Rivul voice run on Rivul AI's own servers. Rivul AI does not train models on your files.

  • Privacy and AI use

Related guides

  • Bring Your Own KeyConnect your own Google Gemini, OpenAI, Anthropic, xAI, Mistral, or Meta API key. See what needs it, and how it is stored.
  • WorkflowsStep-by-step workflows for students, engineers, academics, researchers, and everyday use, each from a source to a result you can check.
  • QuickstartSign in, connect your AI key, run a first search, get a first Ask Rivul answer, run an agent, and connect Rivul MCP.
NextQuickstart

Something here wrong or missing? Use the contact form.

On this page

  • Who it is for
  • From any source
  • Research across it
  • Write, cite and export
  • Free, with your own AI key
  • Where your data goes
Back to top
Rivul AI
Explore
FeaturesDeep ResearchRivul MCP
PricingDocsBlogAbout
Sign in
FeaturesPricingRivul MCPDocsBlogAbout