Tip for private YouTube videos

If videos are private (paid courses), the auto-transcript won't work. Instead: (1) Install the "YouTube Summary with ChatGPT & Claude" Chrome extension, (2) Open the video → click the extension icon → copy transcript, (3) Save as a .txt file and upload it through the Upload tab.

How it works

YouTube links → auto-fetches transcript from captions
Loom links → queued for processing
Website links → scraped and text extracted automatically
Descriptions → used as extra context for AI analysis (better results!)
All links are processed one at a time with rate limiting — your data is never lost.

Tips for best results

• Copy ALL messages from a conversation — more context = better analysis
• Include timestamps if available — helps the AI understand the flow
• "Process as one document" is best for conversations and discussions
• "Split into message batches" is best for long channels with unrelated posts
• You can paste 100+ messages at once — no limit

Quick alternative: Copy-paste method

If you don't want to use Telegram Desktop export, you can:
1. Open the channel in Telegram (desktop or web)
2. Scroll to the top of the messages you want
3. Select all messages (Ctrl+A or click and drag)
4. Copy (Ctrl+C)
5. Go to the Paste chats tab and paste everything there
This is faster but loses some metadata (dates, sender names).

Limitations

• Works best with public pages (blogs, docs, articles)
• Cannot access pages behind login walls (paid courses, private groups)
• JavaScript-heavy sites may have limited text extraction
• For Skool courses behind paywall: copy-paste the text into the "Paste chats" tab instead

Export all (JSON)

Generate full combined report Generate per-category reports

How to upload 1000+ files to Claude (the smart way)

The problem: Claude has a token limit. You CANNOT upload 1000 files at once.

The solution — the Research Engine does the heavy work:
Step 1: Upload ALL 1000+ files to this Research Engine (it processes them over hours, no limit)
Step 2: The engine extracts insights and sorts them into 12 category files
Step 3: Click "Generate per-category reports" above
Step 4: Download each category report (small .md files, ~5-50KB each)
Step 5: Open Claude chat → upload ONE category report at a time
Step 6: Tell Claude: "Analyze this [business/n8n/etc] research for Velstron"
Step 7: Repeat for all 12 categories

Result: Claude reads concentrated, organized intelligence from 1000+ files — not the raw files themselves. This fits in its context window and gives the best analysis.

Upload order (recommended)

1. Business Models — most critical for Velstron revenue strategy
2. AI Agents — core of Velstron's agent army architecture
3. n8n Workflows — Velstron's orchestration engine
4. Claude Code — how Velstron deploys and manages everything
5. Automation Patterns — system reliability and scaling
6. Marketing & Growth — client acquisition for the agency
7. Content & YouTube — Phase 1 revenue engine
8. Tools & Platforms — tech stack decisions
9. Website & Frontend — velstron.com and client sites
10. Claude API — cost optimization for agent operations
11. Trading & Finance — Phase 2 bonus revenue
12. Mistakes & Lessons — what to avoid

After uploading all 12, tell Claude: "Now create the Velstron Blueprint v3.0 based on all research."

Alternative: Add reports to Claude Project files

Instead of uploading in chat, you can add the category reports to your Claude Project:
1. Go to claude.ai → Your Velstron project
2. Click "Project Knowledge" or "Files"
3. Upload all 12 category report files
4. They become permanent context for every conversation in that project
This way Claude ALWAYS has access to your research — no need to re-upload each time.