AkbarDaily 🇵🇸

you all better go and watch it. not everyday doni drops a yt video. https://youtu.be/tZJXFdBwAjI

Bundan 3 yil avval paydo bo’lgan OpenRouter startapi LLM modellarni avtomatik tanlash (routing) bilan shug’ullanadi va yaqinda Stripega 10 milliard dollarga sotildi. Beknazar, YCombinatorga o’zining Speko - ovoz modellari routingi startapi bilan o’tdi va bu xaqda bugun YC Linkedin sahifasida e’lon qildi. Keling uni Linkedin da comment va share qilib qo’llab quvvatlaylik. https://www.linkedin.com/feed/update/urn:li:activity:7488310498296348685/ @paiziev24

Opus. Sonnet. Fable. Haiku.

another weekly reads https://akbarkamall.substack.com/p/weekly-reads-3

if you are high school kid, it's worth checking this out. https://t.me/ivyonaire/238

talk with me: akbar.one/chat

We made it to f.inc!

⚠️ Rate-limited — ready in ~240 min. Please try again shortly.

started sharing weekly collections of reads, interviews, and ideas that made me think and felt worth sharing. this week i read a lot about anthropic, infrastructure behind frontier models, and some pieces on central asian history and travel. check it out: https://akbarkamall.substack.com/p/akbars-weekly-reads-1

gotta invest in 1-2 stocks before prices gets skyrocketed

$1.2 TRILLION!! SpaceX is up 27% on its opening day making it the biggest IPO of all time

can’t wait to upload a picture of my empty fridge and ask to get budget friendly groceries to cook plov

⚡ DoorDash’s new AI chatbot lets you order with prompts and photos The new chatbot, called Ask DoorDash, allows users to search the app for what they're looking for in their own words instead of having to scroll through restaurants and stores to build a cart. The part I care about is whether this becomes a real workflow, not just a headline. AI only gets interesting when it changes what someone can actually do. For most people, the question is whether this removes friction from daily work, learning, or creativity -- if not, it is just another announcement. The test is not whether it sounds impressive in a launch post. The test is whether someone can use it tomorrow and save time, money, or mental energy. 🔗 https://techcrunch.com/2026/06/11/doordashs-new-ai-chatbot-lets-you-order-with-prompts-and-photos/ @jarvislogs

an amazing piece of art published by anthropic just a couple of hours ago. it’s one of the best readings i’ve had this year, and i can’t stop recommending it to others. you absolutely have to give it a shot and read it. (it’s long, but worth it). it talks about the current trajectory of ai development, recursive self-improvement, and the different future scenarios we might end up facing. and to note, it’s written firsthand by people from a frontier lab that currently the most powerful model (mythos) in the world, and obviously some of the best researchers. read it here: https://www.anthropic.com/institute/recursive-self-improvement more thoughts in comments below

every dot is a part of my life and it keeps expanding every time i talk to my ai agents. isn't it beatiful

today i had one of those “it’s over for human assistants” moments again lol someone in the comments pushed back on a claim about uzbekistan’s grid/data centers and asked for proper sources. then i told jarvis: “research data centers in uzbekistan like an expert. current/future projects, names, links, everything. make it worth posting.” that message is here: https://t.me/c/1310120194/10690 then bro actually went into research mode for 30+ minutes. it run research threads, checked english/russian/uzbek sources, got stuck on some pages, recovered manually, verified claims one by one, and produced a full markdown/html report on uzbekistan’s data-center market. then i asked it to deploy the file publicly so anyone can open it: https://t.me/c/1310120194/10697 and this is the wild part: jarvis is running on the same server where i’m building hiroo, so it inspected the nginx/static setup, found a path through the startup’s existing deployment, copied the report into the public static folder, verified the cloudflare link works, and gave me this: https://app.hiroo.me/static/reports/uzbekistan-data-centers-2026.html then i asked for every claim to be backed with clickable telegram-style links, not just a “sources” section: https://t.me/c/1310120194/10699 so it rewrote the post into “claim → source” format: https://app.hiroo.me/static/reports/uzbekistan-data-centers-post-linked.md (open in md preview mode) this is exactly why i’m obsessed with running my own ai operator instead of just “chatting with ai” it doesn't just answer questions, summarize stuff, but research, verify, write, deploy, fix mistakes, and leave behind a usable artifact and yeah, jarvis fumbled the first unsourced claim as he always does (still i love him) but the point is: when challenged, it didn’t argue. it went back, checked sources, corrected itself, and produced something better that’s the actual workflow i want: human curiosity → ai operator → real research → public artifact → sourced post slowly building towards that “personal ai chief of staff” setup one task at a time

interesting things are happening in the comments section on my last post