Grounded RAG • Zero Data Retention • 2-Host Audio

Conversational Research Partner

Upload Locally, Research Privately. Chat with your documents, web pages, and YouTube transcripts with local vector search, inline grounded citations, and viral 2-host audio deep dives. Your data never leaves your browser.

🎬 Official Product Walkthrough (60 Seconds)
crp.dgrvip.net
1
In-Memory Parsing PDF, DOCX, Code & YouTube in client RAM
2
Grounded Citations Inline chips [1] trace claims to text
3
2-Host Audio Pods Rachel & Marcus deep dive podcasts
What is it

A self-hosted research companion that reads your documents and speaks their language.

Local session-based vector search. No database. No cloud storage. No uploads to any external server.

Just your documents, your browser, and the AI — synthesizing answers with grounded citations, 2-host podcast deep dives with Rachel & Marcus, and instant executive briefing documents.

Capabilities

Everything your research needs

Multi-format ingestion

PDF, Word DOCX, Markdown, plain text, web articles, and YouTube video transcripts — all parsed locally in the browser.

Local vector search

Documents are embedded and stored in client RAM. Cosine similarity sublinear search — zero external databases.

2-Host Audio Deep Dives

Generate viral NotebookLM-style podcast overviews hosted by Rachel & Marcus with interactive waveform audio playback.

Grounded citations

Every AI response includes clickable citation chips [1] that trace claims directly back to the exact source chunk.

Privacy by design (ZDR)

Your documents never touch external servers or persistent databases. Ingest proprietary corporate docs with total peace of mind.

Commercial SaaS Ready

Integrated Stripe Pro checkout, plan quotas, 1-Click Studio synthesizers (Briefings, Study Guides, Timelines), and cloud routing.

How it works

Three steps to insight

01

Upload your sources

Drag and drop PDFs, DOCX files, or paste web URLs and YouTube links. The app parses everything locally — nothing is uploaded to any server.

02

Embeddings are computed

Documents are chunked and embedded in-memory. The resulting vectors are indexed for your browser session with zero data retention.

03

Ask & Synthesize

Type a question or click Studio deep dive. The platform synthesises comprehensive answers with citations and 2-host audio overviews.

Get started

Ready to research smarter?

Upload Locally, Research Privately. Launch the workspace directly or subscribe for unlimited Pro reasoning.

Gemini 2.5 WebGL2 Claude 3.5 GPT-4o PDF.js Mammoth.js Stripe Pro 2-Host Audio 100% In-Memory