Shutri: Collaborative Intelligence

Shutri is an open peer-review initiative. Traditionally, peer review has been anonymous and confidential. We are changing that: we run initial reviews using deepMind AI, and our reviewers publish transparent, public reports as proof of understanding to collaboratively learn and spread awareness of original research. You can explore the full architecture and README on the GitHub Repository.

Open-source human-AI peer review registry & research portal.

deepMind@shutri:~$ query-terminal
Ask deepMind
Quick Queries:
Welcome to the deepMind Q&A Ingest Interface. Use the quick tags or enter a query in the prompt below to interrogate the active knowledge database.

Explore the Peer-Reviewed Archive

Shutri operates as a non-commercial, open-access journal for collaborative intelligence. We work as editorial partners to publish peer-reviewed papers, detailed theoretical critiques, and synthesized audio-visual deep dives. Every accepted submission is integrated as an active reference baseline in our permanent vector knowledge engine. You can browse the published volume at deepdive.shutri.com.

▪ Open-Access Publishing: Completely free, open-license, cookie-free, and fully accessible offline via our PWA.
▪ Multimodal Syntheses: Every approved paper receives custom analytical reviews, synthesized deep-dive discussions, and visual summaries.
▪ Permanent Vector Integration: Finalized articles are committed to the deepDive blockchain to serve as active references for future submissions.
Browse Published Volume

The Vision: Hardening AI-Driven Research

In the era of hyper-abundant synthetic intelligence, open collaboration and verified distribution protocols are critical. Here is how Shutri solves the fundamental pain points of modern knowledge networks.

Pain Point: Inbound AI Flood

Block-based Consensus

Instead of an unguided sea of automated papers, we lock and block insights. We group verified articles in sequential blocks of 21 and mint them to our permanent, indexed vector chain.

Pain Point: broken Math / LaTeX

Precision LaTeX Support

Most publishing interfaces render math equations poorly or crash on delimiters. Our ingestion engine utilizes native KaTeX syntax sanitization, ensuring clean rendering of complex proofs.

Pain Point: High Density Disconnect

deepMind Notebook Synapse

We digest the theoretical core of research and synthesize it in Google NotebookLM. This generates a two-host audio conversation embedded directly as a player carousel.

Pain Point: Slow Media Pipelines

On-Device Whisper & DDMA

Our DeepDive Media Automator runs local Whisper models on-device. Reviewers select sub-second word ranges, mix sound stings, and output high-fidelity reels instantly.

Pain Point: Locked Insights

100% Open & Sovereign

All review artifacts, transcripts, sound files, and tools operate under the Creative Commons CC BY-SA 4.0 license. The knowledge substrate remains open to humanity forever.

How It Works

1

Submit Research or Startup Audits Phase 01

Submit your article, blog post, thread, or research paper. We look for mathematical rigor, logical clarity, or unique perspective shifts.

Provide your publication link, name, and email in the portal. Ingestion is completely free. Submissions are staged transparently as GitHub Issues for initial assessment.

2

deepMind AI Ingestion to Mempool mempool

Accepted research is run in Gemini with targeted questions to draft our perspective, extracted via mdIngest, and published to the deepDive Mempool.

Once accepted, we query the research in Gemini, construct our perspective on the submission, extract it using `mdIngest`, and publish it to the **Mempool** on `deepdive.shutri.com`. At this stage, the submission is open and looking for human reviewers to partner with the author to collaboratively create the audio and infographic artifacts.

3

Human Reviewer Partnership & Template Phase template

Once the collaborative audio/video artifacts are complete, the research transitions to the Template phase and is published on global platforms for public review.

The human reviewer works as a partner with the author to finalize the review notes and construct high-fidelity NotebookLM podcasts and Mosaic infographics as proof of understanding. When these artifacts are complete, the research is promoted to the **Template** phase and published across Spotify, Apple Podcasts, YouTube, TikTok, and Instagram for a public feedback window.

4

Consensus & Vector Chain Minting chain

After the public review window closes and 21 finalized articles are compiled, the block is minted to the deepDive blockchain and synced with the vector base.

Once the public review closes and 21 peer-reviewed articles are compiled to form a complete block, they are committed to the **deepDive blockchain** as the next block. The minted notes are integrated directly as active reference baselines in the deepMind vector query database, serving as active benchmarks for future submissions.

Interactive Phase Transition Preview

Submission Portal

Select Ingestion Pipeline

Note: Submissions are hosted transparently as GitHub Issues. You will be prompted to log in to GitHub to finalize publishing.

✓

Submission Staged

Your research link has been successfully logged into the Mempool queue.

[1/5] Validating URL structure... Done.
[2/5] Verifying Creative Commons agreement... Verified.
[3/5] Compiling Markdown issue payload... Done.
[4/5] Preparing routing configurations for GitHub... Done.
[5/5] Launching GitHub Issues portal... Done.