AMERICAN DATA CENTERS Excellent piece in FT by my friend @joshzoff arguing that the US can't treat data centers like we treated rare earths, and that data centers are a great way to finance development of a bunch of other advanced technologies (strong agree).
ECONOMY
-

AI Could Reduce Staffing in Marketing, Finance, and HR
By
–
“Nikesh thinks many G&A functions — marketing, finance, HR — could have half the people over the next three years because AI will make the average employee dramatically more capable.”
-

Enterprise AI ROI Needs Measures Beyond Licenses and Deployment Counts
By
–
AI Reality Check
CEO: "We deployed AI."
Consultant: "What's the ROI?"
CEO: "…"
Many companies measure licenses, users and deployments.
Few measure revenue impact, productivity gains or time saved.
How are you measuring AI success?
#AI #ROI #EnterpriseAI -
Claude Fable Finds and Fixes Five Release Blockers in Review
By
–
Somewhat humbling to have Claude Fable do a final review of some software that you're about to release and have it then find (and fix) FIVE release blockers, for an estimated (unsubsidized) cost of $149.25
-

AI Can Cost More Than the Workers It Was Meant to Replace
By
–
“AI is now costing some companies more than the people it was supposed to replace.”
-
Fable’s Reasoning Settings Take 12 Minutes to Nearly Two Hours
By
–
I ran the same prompt through Fable (in CC) at different reasoning levels, low effort too 12 minutes, max effort took nearly 2 hours.
— Peter Gostev (@petergostev) 4 juillet 2026
Prompt and the hosted version of this video below https://t.co/35jOwP66Ed pic.twitter.com/0xHtStdO5jI ran the same prompt through Fable (in CC) at different reasoning levels, low effort too 12 minutes, max effort took nearly 2 hours. Prompt and the hosted version of this video below
-

Fable Runtime Rises From 12 Minutes to Nearly Two Hours
By
–
Fable – same prompt, different reasoning efforts – 12 mins vs nearly 2 hours between 'low' and 'max'
-
Multi-Agent Coordination Can Add More Cost Than Value
By
–
Glad the coordination point landed, most multi agent setups add collaboration that never earns its cost
-

Sonnet 5 Ranks Thirteenth in a Preliminary Cost and Token Benchmark
By
–

Rumours are true. Sonnet 5 also completely breaks my benchmark in terms of cost and tokens generated. By such a big margin. It's crazily inefficient, and scores 13th behind Fable, Opus, GPT, Gemini, Grok… Voluntarily not sharing more, as this benchmark is still a work in
-
Open Loop Engineering Framework Structures Quant Trading Agents
By
–
This is crazy
— AI Highlight (@AIHighlight) 4 juillet 2026
A free github repo with 4.8K stars just dropped the entire "loop engineering" framework for trading agents.
12 steps to build a self-running quant desk: strategy intent → market data → signals → trading agent → verify → refine → rerun.
Save and bookmark no… https://t.co/01fWpMusQv pic.twitter.com/8yN7cvkt8eThis is crazy A free github repo with 4.8K stars just dropped the entire "loop engineering" framework for trading agents. 12 steps to build a self-running quant desk: strategy intent → market data → signals → trading agent → verify → refine → rerun. Save and bookmark no
