you are my lead and the only bot i talk to. team: scout, coder, metrics, reviewer. at 07:30, after the reviewer, write my briefing: 1. decisions for me today (max 3), with the draft waiting 2. what moved in my numbers (reviewer-passed only) 3. what's new in my niche (reviewer-passed only) 4. what failed overnight and why one screen. no item without a link. during the day, route my tasks to the bot that owns them. hand over the goal, the context, and why, not just the file. if no bot owns it, tell me.

CLAUDE CAN NOW SPEND YOUR MONEY stripe just plugged into claude duet, the proactive agent anthropic opened yesterday -> with link it can order product samples, send gifts or top up office supplies -> with the stripe MCP it can issue refunds and pull your business trends the crazy part is the guardrail. you approve every single purchase, and your payment details stay private to be fair duet is still a waitlist research preview, pro and max in the US only, so most people won't touch this for a while personally refunds are the part i'd try first. a customer asks, duet finds the order in stripe, asks me once and it's done

ANTHROPIC JUST KILLED OPENAI'S DOTS anthropic just opened claude duet, a new app where claude works as a proactive agent -> it comes to you with what needs your attention -> it finishes the tasks you hand it -> it keeps working when you walk away -> it checks with you before it acts on your behalf it runs on opus 5.5 and pulls in your memory, skills and connected apps from your claude account, so there's almost nothing to set up the crazy part is the safety layer. it never directly sees your passwords, and a new classifier guards against it sharing anything sensitive to be fair dots is live and duet is still a waitlist research preview, pro and max in the US only. anthropic even says it can still misjudge what's urgent personally i'm joining the waitlist tonight. the first thing i want to see is what it decides needs me meanwhile at openai HQ:

Meta's AI chief, Alexandr Wang: "we actually delayed the launch of Muse for many months to get the safety and security exactly right" in 68 minutes he explains what it takes to trust an AI agent with your personal info, your phone calls and your tasks, and why meta built a separate sentinel agent just to watch what muse sends out your own agent needs the same thing on a smaller scale. the article below shows the one file that keeps it in line while you sleep

GOOGLE JUST DROPPED GEMINI 4 its called gemini 4 argon, and google says its #1 on the vals index, a benchmark built on real work in finance, coding, law and tax what google says it already did inside google: -> helped the chrome team untangle a 6 million line chunk of old code. months of work done in weeks -> made their open gemma model almost 3x faster in a few days. normally that takes months -> their engineers using it shipped 35% more code with almost a third fewer rollbacks the price: $2 per million input tokens, $10 output. exactly what anthropic charges for sonnet 5.5 who gets it today: developers on the paid API, plus pro and ultra subscribers in the gemini app and AI mode in google search the crazy part: openai shipped gpt-6, anthropic shipped haiku 5.5, now google answers with argon. 3 labs in one week to be fair these are google's own numbers, and free users dont get it yet personally the price is the real headline. a frontier model at sonnet prices means the "best model" race just turned into a price war

MICROSOFT'S CEO JUST CALLED AI AN INSIDER THREAT satya nadella just published an article everyone running AI should read. 9 million views in a day his whole point in one line: "we need to separate the supply of intelligence from the authority over it" what he says every company now needs: -> never let one model check its own work -> every action the AI takes leaves a record a human can read. "If it can't be observed, it can't be trusted!" -> the company decides what the AI can touch, not the AI -> assume the model is compromised from day one. a human can always hit the brake mid task -> when it fails, tell the people it affected his last line: "The most trustworthy Super Intelligence system will not be the one with the model we trust most. It will be the one that enables us to trust the model the least." the crazy part: this is the CEO of one of the biggest AI sellers on the planet, telling his own customers not to trust the models personally i built the tiny version of this in june. a rulebook my AI reads before it works overnight: what it can do alone, what it has to ask me, what it can never do. same idea, one file. link in the replies

CLAUDE JUST CAME FOR YOUR WHOLE OFFICE anthropic just dropped two new things and made three more free for everyone claude dashboards: -> connect your company data. bigquery, snowflake, databricks, salesforce -> ask a question in plain english -> claude writes the query and builds a dashboard that keeps updating as the data changes -> every chart shows the query behind it, so you can check it claude motion: -> turns reports, charts and product walkthroughs into short animations -> no video model involved. claude writes each animation as code, so you can change any word, number or timing and export an MP4 and claude docs, slides and design are out of beta and on every plan, including free the crazy part: until now a question about company data meant filing a ticket or writing SQL and waiting. now its one sentence to be fair dashboards is beta on paid plans, and motion is beta only on team and enterprise personally this is the same week claude moved into google docs. anthropic isnt building a chatbot anymore, its building the whole office around it

A MATH PROFESSOR JUST RATED OPENAI'S 722 PAPERS abhishek saha, a maths professor in london, sorted big math results into 4 levels: -> A: solid normal results. most of published math lives here -> B: exceptional advances. top journal level -> C: surprising breakthroughs that change a whole field -> D: shock breakthroughs. experts thought they were basically impossible right now. "instant Fields medal variety" the best mathematicians prove only a handful of B and C results in their whole life. many never prove one his verdict on openai's drop: some are A, but most are B or C. and exactly one is D, the one people are calling the quasi riemann hypothesis his last line: "It is a very big day for mathematics." the scariest part: all of it came from one unreleased model, using about 3 hours of thinking per result to be fair nothing is fully confirmed yet. openai itself says some proofs could still have issues, and checking all of it will take mathematicians months personally this is where i stop calling AI a tool. a tool doesnt do in one release what the best humans do in a lifetime

CHATGPT WILL START HIDING AN INVISIBLE MARK IN EVERYTHING IT WRITES FOR YOU openai just confirmed it. its called textGrain and it hits chatgpt and codex in the EU over the next few weeks, because the EU AI Act forces it you wont see anything. no weird symbols, no hidden characters. the model just picks its words in a slightly patterned way, and a detector can read that pattern later what it can tell: "was this likely written by an openai model" what it cant tell: who you are, your account, your prompt the funny part is openai itself says it barely holds up: -> short paragraph: caught about 80% of the time -> change 1 in 10 words: detection drops from 92% to 66% -> change 1 in 4 words: 17% -> translate it or rewrite it: the mark can disappear completely the scariest part isnt the watermark. its who actually gets caught anyone who edits a little walks free. the kid who copy pastes a full essay without touching a word is the one it catches. and only "approved researchers" get the detector for now, so you cant even check your own text to be fair, openai isnt hiding any of this. they published the weak numbers themselves, API users anywhere can switch it on today, and they say theyll open source it personally i think this is bigger than a watermark. europe gets a chatgpt that marks its text, america doesnt. same AI, different rules depending on where you open it and this is just the first law. there will be more

TRUMP JUST GOT EVERY MAJOR AI CEO TO SIGN A PROMISE TO POLICE THEMSELVES. HE CALLS IT "MORALLY BINDING" yesterday at the white house: musk, zuckerberg, huang, amodei, pichai and brockman signed the "White House Accord on Super Intelligence" one page. four commitments: internal controls on cyber, bio and chemical risks internal teams to check the controls work independent outside auditors board committees to review the reports penalties if they break it: none. its voluntary trump: "It's almost like a constitution, in a way" "I think it's morally binding" 6 days earlier bernie sanders filed a bill with up to 20 years in prison for building superintelligence. this week we got a signed promise to be fair, outside auditors and board level oversight are real steps. its more than most of crypto ever agreed to but i've heard "we have independent auditors" before FTX had outside auditors. two firms. both signed off on its books. it still went to zero the scariest part isnt that the accord is weak. its that the auditors get hired by the same companies theyre supposed to check personally, crypto taught me one thing about self regulation. it works right up until the first quarter where it costs money so which lab quietly misses a commitment first, and would we even find out?