knowledge-gate: AI talks into my own notes, while I sleep

Description and use

The system automatically processes new AI talks, verifies quotes, builds a concept map, and updates local notes with only new content, saving time and ensuring consistency.

Problem

AI engineering talks are released faster than the user can watch them, making it hard to capture key ideas manually.

Outcome

The system automatically ingests new AI talks, verifies quotes, builds a concept map, and updates local notes with only new content, saving the author time and ensuring consistency.

Workflow

• 1. YouTube push notification triggers webhook
• 2. Temporal workflow starts
• 3. Transcript tool extracts transcript
• 4. Card Writer (Gemma 4 31B) writes a one‑line conclusion, summary, and claims
• 5. Quote Checker verifies word‑for‑word quotes
• 6. Concept Matcher (Gemma 4 31B + bge‑m3) links to existing concepts or creates new ones
• 7. Category Keeper files the talk under a section
• 8. Map step updates the concept map
• 9. Daily laptop task pulls new cards
• 10. Note Merger (Gemma 4 12B + EmbeddingGemma) appends only new points to local notes
• 11. Notes Assistant (Gemma 4 12B + EmbeddingGemma) answers queries from notes and talks

Evidence

When a talk is uploaded, a small server turns it into a card: a one-line conclusion, a short summary, and claims, each with a word-for-word quote and the second it was said.

Source

The cards grow into one map of concepts, so ten talks about one idea become one page.

Source

Every morning, when I open my laptop, a local model adds only what is new to my notes.

Source

The Quote Checker keeps a quote only if it is in the transcript word for word.

Source

The merge only appends. After every merge a git check reports any existing line that was removed or changed.

Source