A research brain for scientific research

Meet Rao: an always-on, proactive scientific collaborator that lives across the surfaces where you work, powered by research memory that learns new project knowledge as you do.

Helping researchers at leading institutions discover more, faster
Harvard Northwestern MIT EPFL NYU UC Berkeley

AI that remembers every scientific detail.

Rao connects your project and apps into one research brain that powers its agents and your tools. Rao Code & scripts Notebooks Papers & PDFs Data files Past AI sessions CONNECTED APPS Notion GitHub Google Docs Google Sheets RAO AGENTS Proactive ideas Report generation Subagent orchestration YOUR TOOLS, SMARTER Claude Code Codex Cursor Claude Science Code & scripts Notebooks Papers & PDFs Data files Past AI sessions CONNECTED APPS Notion GitHub Google Docs Google Sheets Rao RAO AGENTS Proactive ideas Report generation Subagent orchestration YOUR TOOLS, SMARTER Claude Code Codex Cursor Claude Science

A team of proactive research agents.

Proactive ideas
arXiv ChemRxiv Nat. Mater. Phys. Rev.
Your open question
RaoAPP8:30 AM

A new strain-relaxation model could explain the conductivity drop in your 400 °C anneals.

Launch subagent
Report generation
Weekly ensemble reportOct 12
02_ensembles.ipynb
Notion · Sep 16
GitHub · #42
SlidesDocsMarkdown
Subagent orchestration
RaoIs the 2.3σ excess real?
  • Search literature
  • Refit with a new background model
  • Split by run period
  • Redraw Fig 3
One shared research memory

Rao's brain, inside the agents you already use.

Link Rao into:

Claude Code Codex Cursor Claude Science
Without context Your AI detrend_lc.py kim2026.pdf fig3.png 02_tess.ipynb targets.csv obs_log.xlsx
Only sees saved files and has to re‑infer connections.
With Rao's brain Your AI Kim et al. 2026 came up in Group meeting prompted detrend_lc.py produced Fig 3 Rao research brain
Automatically remembers every detail and their relations.

Get started in 30 seconds.

01

Index your project in minutes

Open Rao and point it at a project folder. It automatically builds a knowledge graph of how your files relate, both scientifically and structurally.

02

Collaborate with context-aware tools

Talk to Rao as you would a close collaborator, and plug Rao's context into other agentic tools via MCP.

03

Build research memory as you go

Rao learns new results and connections automatically, helping you discover more, faster.

Frequently asked questions.

Rao builds a persistent knowledge base for your research. It understands the context and direction of your project - scripts, data, notebooks, PDFs, meeting notes, planning documents - from start to finish, improving the consistency, accuracy, and insights of AI agents. Rao comes with several agents specifically geared towards scientific research (e.g. chat agent, writing assistant, skill builder). You can also improve other agentic tools you're already using by plugging in Rao's research memory via MCP.
By default, the only information transmitted to us is basic usage metadata and the data required to authenticate your account. The contents of your research and your interactions with Rao remain private and local to your machine unless you expressly share them with us.

Separately, because Rao runs on external providers' models, your chat interactions pass through their APIs and are governed by their commercial data retention policies. Reach out to hello@tuva.ai for more details. If you work in a regulated environment — HIPAA, ITAR, IRB-restricted data, or under an industry NDA — check with your PI or management before using Rao to confirm compliance.
Join our waitlist and we'll reach out to arrange an onboarding call when space is available. On the call, we'll walk you through download, installation, and indexing your first project.
Interact with Rao as you would a research collaborator. Rao can both do tasks (write code, do math, search, write), but also suggest research directions, review your work, and explain the provenance and evolution of your files and ideas. It handles both open-ended exploration and specific questions about your project's details, and can build reusable skills for your frequently-used research practices, all grounded in your project knowledge and history.

Want to try it out?