☀️ Today
☀️ Today — your daily orchestrator
Pick a news/event → turn it into an operator take in Composer.
Every morning at 06:00 Asia/Ho_Chi_Minh the system fetches 14 DTC subscription news sources,
scans them with a research model (anchored 0/2/4/6/8/10 relevance scale for $1M–$8M Shopify subscription brands),
and surfaces the top 5 here + sends to Telegram. Click Use as seed on any item to open Composer
with the news pre-loaded; click ⟳ Refresh digest to fetch fresh news on demand.
✍️ Composer
1
2
3
4
5
How this works
Pick what you are starting from, paste the concrete angle or proof, preview the draft plan, then generate.
1. Starting point
2. Draft plan
3. Generate + tighten
1
Format
— what are you creating?
2
Starting point
— what do you already have?
3
Brief + Plan
— paste the brief, preview your plan, generate
Brief formula: topic + audience + frame + specific claim or metric + what to avoid.
Example: “Make a carousel for Shopify subscription operators about Recharge acquiring Skio. Frame: Audit Map.
Angle: platform dependency is now a rebill / dunning / portal / data-ownership risk, not a SaaS news item.
Avoid generic M&A commentary.”
Brief
Paste messy notes. Specific beats polished.
Use an example
Advanced options
Pillar is optional. Use it only when you want the generator to stay inside one strategic bucket.
4
Generate + Edit
— attach an image or skip
📁 Drag & drop or click to upload (multiple OK)
PNG / JPG / WebP / GIF · max 10 MB each
5
Preview + Save
— final review
⚠ Text changed — recompute Voice score / Hook variants
Before saving: check the hook, remove generic claims, shorten overflowing carousel headlines, then run
Score voice. For carousels, open the debug trace if the score feels weak.
Click "Score voice ▶" to analyse the draft.
Click "Get hook variants ▶" to see 5 alternatives.
No assets attached
Loading…
📊 Analytics
Performance across published LinkedIn posts. KPIs + per-post table + pillar aggregates + hook A/B + format breakdown. Engagement funnel (impressions → DMs) activates once comment + DM thread tracking is wired up.
📚 Library
🎙️ Voice Document — the writing rules Composer must follow
This is not the topic knowledge base. It is the voice layer: how Alex writes, what he
never says, which proof patterns he trusts, and what operator language sounds natural. Sonnet turns your
raw notes into a structured Voice Document. The active version is then mixed into Composer prompts so posts,
carousels, comments, and scores can judge against the same style rules.
1 · Purpose
What this controls
Sentence rhythm, hooks, banned phrases, CTA style, proof standards, and vocabulary by pillar.
2 · Input
What you paste
A raw writeup: cases, frameworks, contrarian takes, signature phrases, and words you hate.
3 · Build
What AI does
Converts your notes into machine-readable JSON. Every build creates a new version; only one is active.
4 · Safety
How to rollback
If a new version writes worse, activate an older one from Version History. No content is deleted.
What to fill in
- Anchor frame: who you are, who you serve, and what you have earned the right to say.
- Real cases: anonymized brand size, vertical, symptom, diagnosis, intervention, result.
- Named frameworks: reusable lenses like Month 2 Cliff or Involuntary Churn Recovery Map.
- Contrarian takes: what most operators believe, what you believe instead, and why.
- Signature phrases: phrases you actually use and phrases that should be banned.
- Pillar vocabulary: how you talk about payments, retention, traffic quality, offers, refunds, etc.
- Anchor stats: numbers or thresholds you would defend publicly or use in audits.
- Voice laws: hard rules such as no soft CTA, no generic SaaS language, no fake certainty.
Good input shape:
Brand descriptor: $3M supplements brand on Recharge.
Symptom: founder thought email retention was weak.
Diagnosis: M2 churn was mostly failed payments and cadence mismatch.
Intervention: changed dunning timing, rebill reminders, and subscription cadence.
Result: recovered X% of failed rebills over Y weeks.
Teachable bit: this was billing ops, not lifecycle copy.
New Voice Document Version
Paste a full writeup below. Short notes work for a bootstrap version, but a strong version needs real material:
5-8 anonymized cases, 3-5 frameworks, banned phrases, and concrete operator vocabulary. Expect 60-120 minutes
for a serious version.
Active Voice Document
No active version yet. Build one above.
Version History
| Version | Created | Status | Notes | |
|---|---|---|---|---|
| No versions yet | ||||