Back to the brief

Analytics

Engagement and cost

Daily engagement, last 14 days

07-24
67% of 21 read · 25 useful · 5 not · 2 persp · 14 chat
07-23
114% of 22 read · 33 useful · 10 not · 18 persp · 6 chat
07-22
47% of 17 read · 3 useful · 0 not · 5 persp · 0 chat

Target: engage at least 50% of brief stories daily (PRD metric 1).

By lens, last 14 days

LensUsefulNot for mePerspectives
AI + Tech18214
Geopolitics506
Global headlines1220
Global402
country-cn600
Trending840
country-ru400
AI + Society012
Work011
country-us220
country-ph230

Most engaged stories

Kimi K3 tops coding benchmarks, trails only Fable 5 and GPT-5.6

AI + Tech · 1 useful · 0 not · 4 persp · 2 chat

OpenAI models hacked, exposing AI arms-race risk

AI + Tech · 2 useful · 0 not · 3 persp · 1 chat

Senate Subpoenas Duterte's Financial Records as Impeachment Trial Escalates

Global · 2 useful · 0 not · 2 persp · 0 chat

OpenAI confirms its own AI model hacked Hugging Face in test

AI + Tech · 1 useful · 0 not · 3 persp · 0 chat

OpenAI's accidental cyberattack on Hugging Face alarms security researchers

AI + Tech · 1 useful · 0 not · 2 persp · 0 chat

China runs two-day live-fire drills in Taiwan Strait

Geopolitics · 1 useful · 0 not · 2 persp · 0 chat

Alibaba merges three agent products as big tech reshuffles AI bets

AI + Tech · 2 useful · 1 not · 0 persp · 1 chat

AI Job Displacement Debate Intensifies as Roles Get Automated

AI + Society · 0 useful · 0 not · 2 persp · 0 chat

Houthis claim Red Sea tanker strikes as US-Iran war escalates

Geopolitics · 0 useful · 0 not · 2 persp · 0 chat

Trump adviser accuses China's Moonshot AI of stealing from Anthropic

AI + Tech · 1 useful · 0 not · 0 persp · 1 chat

Houthis Attack Saudi Tankers as US Strikes Iran

Global headlines · 2 useful · 0 not · 0 persp · 0 chat

Russian Forces Advance in Eastern Ukraine Amid Intense Fighting

Global headlines · 2 useful · 0 not · 0 persp · 0 chat

Moonshot AI Accused of Theft; OpenAI Rogue Models Breach Firm

Global headlines · 2 useful · 0 not · 0 persp · 0 chat

Anthropic weighs special employee stock-sale rules ahead of IPO

AI + Tech · 1 useful · 0 not · 1 persp · 0 chat

Kimi K3 Distillation Row Erupts as Microsoft Tests It for Copilot

AI + Tech · 1 useful · 0 not · 1 persp · 0 chat

Most clicked, last 14 days

Every link and button, sitewide. Per-user breakdowns arrive with accounts.

3 Useful /
2 Geopolitics: China Fires Water Cannon at Philippine Ships Near Ayungin Shoal. Flip for why it matters. /welcome
2 To the brief /pass
2 gallery-cta /gallery
1 cn /
1 US and China Escalate AI and Technology /
1 Get new news /
1 Archive /
1 Play today's edition /welcome
1 AI + Tech: OpenAI's AI Agent Autonomously Hacked Hugging Face, Chinese Model Aided Cleanup. Flip for why it matters. /welcome
1 AI + Tech: Claude Voice Mode Expands to Opus and Sonnet Models. Flip for why it matters. /welcome

LLM spend

Estimated $1.41 all time · target under $20/month (PRD metric 2)

PurposeCallsEst costAvg time
arlo10$0.0619.4s
coverage8$0.03134.5s
perspective-cn8$0.09614.9s
pick-morning8$0.117126.9s
write-morning8$0.22636.6s
chat6$0.16913.1s
learn6$0.06950.5s
section-whyme6$0.10659.9s
trending5$0.01327.1s
wire5$0.040193.7s
cleo4$0.05117.8s
perspectives4$0.19038.5s
art-concepts3$0.00520.7s
pick-midday3$0.02786.1s
write-midday3$0.03822.3s
perspective-fr2$0.03110.0s
perspective-ph2$0.0299.2s
perspective-ru2$0.02212.3s
perspective-uk1$0.0139.9s
perspective-us1$0.0149.1s
pick-delta1$0.010134.8s
pick-evening1$0.01059.2s
podcast-script1$0.00819.3s
write-delta1$0.01529.2s
write-evening1$0.01320.3s

What the system has learned

Generated by the engagement analyst before each morning brief; injected into story ranking alongside profile.md.

- Prioritize AI security/hacking incidents (OpenAI↔Hugging Face breach cluster): across 4+ stories, up to 26 scroll, 3 perspective taps, 2 saves, 2 chats on single stories — the single strongest cluster in the data. - Prioritize Chinese AI lab competitive/funding stories (Kimi K3, DeepSeek, Alibaba, Moonshot): 4 perspective taps + 2 chats on Kimi K3, 2 saves on Alibaba, 1 save on DeepSeek; reinforced by reader's own follow-up questions on RMB funding comparisons and the Moonshot-accusation advisor. - Prioritize saves-worthy AI stories regardless of lens tag: 7 of 8 save events sit on AI-related stories (ai-tech, country-us AI policy, country-cn AI framing) vs. 1 on a wire story — saves are the strongest signal and concentrate almost entirely on AI content. - Prioritize China-Taiwan Strait military stories: both instances (geopolitics + country-cn) drew perspective taps and 8 scroll each, and match the reader's explicit "how far is Fujian from the Philippines" question. - Prioritize substantive PH accountability stories (Lacson ghost-projects probe, Hontiveros death-threats, Senate subpoena of financial records): each got useful votes with zero downvotes, n=3. - Deprioritize PH impeachment trivia/polling sidebars (Pulse Asia poll, VP defense-team food costs, "7/10 Filipinos want law-based decision"): all 3 got not_useful with no offsetting engagement — consistent pattern, not a single-story fluke. - Deprioritize celebrity/entertainment items in the trending lens (Model Scout, Powerball, Halo Campaign, Jack Smith): 4/4 not_useful, vs. science/legal trending items (Meteorite Crater, Nintendo Palworld) both useful — keep trending to hard news/tech, cut tabloid picks. - Deprioritize Red Sea/Houthis conflict framed as geopolitics analysis: the geopolitics-lens story got 0 useful despite 8 scroll + 2 perspectives, while the wire-lens Houthis brief got 2 useful + 1 save (n=2, treat as lens-format signal, not topic ban). - Deprioritize kinetic military stories with no AI/fintech/economic angle (US B-1 bomber deployment: 1 not_useful, 0 other engagement) — Iran/military stories tied to tariffs or markets performed better. - Ground why_me on AI-security and Chinese-lab stories in fraud-detection/weaponization risk for PH and Cambodian fintech — directly matches the reader's stated work-stress-test question and should be the throughline, not a generic AI-industry note.