Get high on information.

Drop 500 videos. AI extracts, scores, and organizes the best content. Export to Obsidian, PDF, or any format.

Local-first processing
Your API keys
Zero cloud storage

Xtract. Transcribe.
Compose.

From raw source to structured knowledge in three steps.

X

Xtract

Pull media from 8 platforms. YouTube, Instagram, TikTok, Spotify, Vimeo, X. Entire channels. Entire playlists. 100+ URLs at once.

T

Transcribe

Real transcription via Whisper or AssemblyAI. Speaker diarization. 90+ languages. Not YouTube captions, actual audio-to-text.

C

Compose

AI pulls exactly what matters: quotes, frameworks, contradictions, teaching moments. 250+ prompts tuned per profession. Organize, select, export.

NotebookLM does one video.
You do fifty.

Google's NotebookLM is great for a single source. But it can't bulk-process, can't do custom extraction, can't export structured data, and can't handle 8 platforms. When you need to go from 50 interviews to organized, themed, exportable knowledge, that's a different tool.

See the full comparison
Sources at once 1 50+
Platforms Google Docs, PDF 8 platforms
Extraction modes Summary only 250+ prompts
Export Copy-paste 6 formats
Data ownership Google's servers Your machine

The full pipeline

Eight steps from raw source to finished output. No other tool does this.

Extract

Paste URLs from 8 platforms. Channels, playlists, 100+ URLs at once.

Transcribe

4 engines, speaker diarization, 90+ languages.

Correct

AI-powered transcript cleanup. Rename speakers, fix errors, waveform editor.

Subtract

Two modes: Preserve exact words (for books) or Distill ideas (for research). 250+ prompts across 10 professions. Or let AI design a custom prompt for your goal.

Organize

AI scores every block 1-5, suggests Keep/Star/Trash. One click to approve. Finds duplicates across sources. Semantic clustering groups content by theme.

Select

Fine curation. AI compares similar pieces side by side, explains which is stronger and why. Star the best of the best.

Compose

Assemble final output: book chapters, course modules, evidence tables, contradiction matrices. Whatever your deliverable needs to be.

Publish

Export to Obsidian (wikilinks, frontmatter, tags), PDF, DOCX, JSON, Markdown, TXT. One click to your knowledge vault.

Paste 100 URLs. Go make coffee.

  • Bulk import entire YouTube channels, playlists, or URL lists
  • One-click batch download, transcription, and extraction
  • XTC processes everything in the background while you work on other things
  • Come back to hundreds of organized, extracted blocks ready for human curation

The AI does the heavy lifting. You make the creative decisions.

AI does the work. You make the decisions.

Not a summarizer. An intelligent content platform that understands what you need.

AI Guide

"I have 200 depositions, need contradictions." The AI reads your capabilities, asks questions, and plans your entire workflow. Step by step. No learning curve.

Custom Prompt Engine

AI scans your content ($0.02), then designs a precision prompt for your exact deliverable. Contradiction matrices. Evidence tables. Course curricula. Not generic summaries.

AI Curation

Scores every block 1-5. Suggests Keep, Star, or Trash. Explains reasoning. Finds duplicates across sources. You review and approve in one click.

Talk to XTC Studio. It plans everything.

No manual required. Describe what you have, what you need. The AI Guide does the rest.

1
You say
"I have 400 videos of a spiritual teacher. I need a photo book with 150 quotes and a full book of dialogues organized by theme."
2
AI Guide asks
"What language are the videos in? Does the speaker have a distinctive voice you want to preserve? Are the quotes short (1-2 sentences) or longer sayings (3-7 sentences)?"
3
You answer
"English, French accent, her exact words must be preserved. Mix of both short and long."
4
AI Guide delivers your plan
Your workflow plan:
Step 1 Extract tab: Import all 400 YouTube URLs
Step 2 Transcribe tab: AssemblyAI with speaker diarization ON
Step 3 Correct tab: AI identifies "Speaker 1 = Ganga Mira"
Step 4 Subtract tab: Preserve mode, Author style, Opus model
Step 5 Organize tab: AI scores all blocks, approve triage
Step 6 Select tab: Star 150 for photo book, 300 for Fierce Grace
Step 7 Publish tab: Export photo book as PDF, Fierce Grace to Obsidian
Estimated cost: ~$12 for Opus on 400 videos. Time: transcription runs overnight, curation takes 30 minutes.

Works for any goal. "I need a course from 100 expert interviews." "Find contradictions in 30 depositions." "Build a knowledge base from 50 lectures." The AI Guide reads your content and plans everything.

250+ extraction prompts. Tuned per profession.

The Style Library: 10 profession-specific extraction lenses, 25 prompts each. Every prompt designed for how that profession actually works.

Style Library

Researcher, Student, Legal, Medical, Psychology, Filmmaker, Course Creator, Influencer, Journalist, Author. Each with prompts built for their workflow.

Custom prompts

Write your own extraction logic. Refine iteratively until the output matches what you need. Save as presets for reuse.

Cost preview

See the estimated LLM cost before you run any extraction. No surprises on your API bill.

See all prompts

Built different

Your profession. Your extraction lens.

Each style extracts differently because each profession needs different things from the same source material.

See all 10 profession styles

Summaries throw away 98% of your source material.

XTC extracts the actual content. Every quote, every framework, every contradiction.

Feature XTCStudio NotebookLM Otter.ai
Bulk processing (50+ sources) ~
Real transcription (Whisper)
Custom extraction per profession
Export to Obsidian / JSON / MD ~ txt
Data stays on your machine
See full comparison

Stay tuned.

Join the waitlist for early access.