Run your copywriting system like a studio.

Ingest material, route it through the right knowledge bases, and send different assistants to generate platform-specific outputs.

Materials this week42
Knowledge bases9
Active assistants6
Reusable outputs128

Material intake

Every source enters here first. Links, files, and text all converge into a transcript-centered asset.

Paste a Douyin or YouTube link here. Knowledge target: Founder Hooks + Proof Framing Assistant route: AI Startup Scriptwriter + Sales CTA Finisher Analysis model: Local GLM-5
VC founder talking-head sample 12m · transcript ready · hook density 8.9/10
Linked
High-conversion CTA collection text note · 38 CTA endings extracted
Text

Knowledge routing

Make the knowledge graph visible. Users should always know which assistant can read which material pool.

1. Transcript clean-up Remove filler, split hooks, isolate claims, normalize CTA language.
2. Style abstraction Extract rhythm, sentence energy, authority level, objection handling patterns.
3. Knowledge base sync Founder Hooks, Sales Emotion, Short CTA, Proof Framing.
4. Assistant generation Bind one or more KBs, assign model, generate title/script/variant bundle.

Output assets

One material source should fan out into multiple reusable content assets immediately.

AI Startup Scriptwriter Bound to Founder Hooks + Proof Framing · Model: Local GLM-5
60s oral script 3 title variants Closing CTA
Emotion-driven Sales Closer Bound to Sales Emotion + Short CTA · Model: Gemini via local proxy
Private-domain follow-up Urgency rewrite
Authority-led Brand Explainer Bound to Proof Framing only · Safer for educational content
Long caption Carousel outline