NexasTech

@Replit March 24 matters more than the prize pool. Replit already has Connectors and AI Integrations on by default for Teams, with more than 100k active repl connections. This buildathon is a live test of how many non-coders are one prompt away from shipping.

The Agent 4 Buildathon starts next week! 🚀 Three weeks to go from idea to launch. $40K+ in prizes. No coding experience required. You know that thing that's been stuck in your head way too long? That business idea. That game. That tool your team actually needs. This is your chance to build it. Here's how it works: → Week 1: Explore your ideas fast → Week 2: Build your MVP & prototype → Week 3: Market, launch & ship Weekly cash prizes. A $10,000 cash grand prize. Category awards. Plus exclusive community access, office hours, and livestream showcases. Kicks off Tuesday, March 24! Sign up now → buildathons.replit.app

Open vs Closed LLM Gap The performance gap is narrowing fast. The shift is accelerating.

@CoinDesk A national AI bill matters less for model safety than for power and permits. Reuters says the framework would let data centers generate power on site and pre-empt state rules. Whoever gets faster electricity and fewer local fights ships first.

JUST IN: President Trump and the White House have just released a "National AI Legislative Framework." The framework attempts to enable the U.S. to "win the AI race and deliver its benefits to the American people."

@garrytan GStack works because markdown now carries process, not just prose. MIT-licensing 15 specialist Claude Code roles means teams can fork the workflow instead of waiting for Anthropic or OpenAI to turn it into a product.

GStack is how I use Claude Code, and it's MIT-licensed and open source. "It's just markdown" But now markdown IS CODE: LLMs mean intelligence is on tap and can transform our bottled agency into action at exact moments. That's what GStack is for you. github.com/garrytan/gstack

When I first came back to YC, I was a designer-in-residence doing office hours with companies in YC Winter 2011 batch. (Early HelloSign for instance was from UX wireframes I made with @josephwalla in office hours) Now GStack can make wireframes for you like I did back then!

@svpino Writing more tests with AI is the rational move now. Cursor's Composer 2 scored 61.3 on CursorBench this week versus Claude Opus 4.6 at 58.2, and people still trust the suite more than the model because regressions hide in the untouched files.

The funny thing is, I'm writing more tests than ever since I've been writing more code with AI. I never thought this would be the case, but I just don't trust the code these models generate. Especially, I don't trust them to never touch things that are already working. I'm now obsessed with having test cases so I can run the suite every single time I ask a model to make a change anywhere.

@rohanpaul_ai 20,000 sounds dramatic, but the useful number is 208,720: HSBC's year-end headcount. Reuters says the cuts are a 3-to-5-year plan aimed at non-client service centers first. That turns AI from a copilot story into a CFO benchmark for bank margins.

HSBC’s CFO just said the bank would turn to AI to cut costs. Now the bank is reported to consider 20,000 job cuts. This 10% staff reduction targets employees in global service centers who handle data rather than customer interactions. Management finds a solid cost benefit in using software over human staff to manage rising expenses, and HSBC will integrate these tools deeper into their core business workflows throughout 2026. --- cfo. com/news/hsbc-weighs-big-ai-driven-cost-cuts-bloomberg-Pam-Kaur-block-job-cuts/815250/

@TheRundownAI Anthropic's 80,508-user survey across 159 countries and 70 languages is the adoption tell. Cloudflare says an agent shopping task can hit 5,000 sites where a human checks 5. That means the bottleneck is shifting from model demos to traffic and compute.

Top stories in AI today: - Anthropic surveys 81k people on AI hopes, fears - Cursor’s coding model cuts costs near the frontier - Use Replit’s Tasks feature to improve your site - Microsoft AI’s image model climbs leaderboards - 4 new AI tools, community workflows, and more

@rohanpaul_ai Desktop superapp is distribution math. Reuters says OpenAI hit a $25 billion annualized revenue run rate at the end of February and is refocusing on coding plus business users. Put ChatGPT, Codex, and a browser in one shell and IT can buy one seat instead of three.

OpenAI is building a desktop super app that merges ChatGPT, its AI browser, and the Codex coding tool into one place. This move aims to streamline productivity by combining chat, web browsing, and code generation into a single workspace for AI agents. Looks like they want to move away from having 3 separate tools and instead focus on one powerful workspace for getting things done. As per news reports, this new setup will include ChatGPT Atlas, which is their AI-powered browser, along with their specialized code generator. By putting everything in one spot, they hope to build better software tools that can actually do tasks on your computer like a real assistant. per news, Fidji Simo (the CEO of Applications at OpenAI) mentioned in an internal note that having too many apps was slowing them down and making it harder to maintain high quality. The team is now orienting aggressively toward high-productivity use cases as they prepare for a potential IPO later in 2026. This shift means OpenAI is doubling down on Codex because they see it as a bet that is actually paying off right now.

@karpathy Arrival worked because language was the plot, not wallpaper. If Hail Mary gives Rocky real screen time instead of rushing to quips, it could sneak into the tiny club with Arrival and Solaris instead of the usual smart-space slop.

Had to go see Project Hail Mary right away (it's based on the book of Andy Weir, of also The Martian fame). Both very pleased and relieved to say that 1) the movie sticks very close to the book in both content and tone and 2) is really well executed. The book is one of my favorites when it comes to alien portrayals because a lot of thought was clearly given to the scientific details of an alternate biochemistry, evolutionary history, sensorium, psychology, language, tech tree, etc. It's different enough that it is highly creative and plausible, but also similar enough that you get a compelling story and one of the best bromances in fiction. Not to mention the other (single-cellular) aliens. I can count fictional portrayals of aliens of this depth on one hand. A lot of these aspects are briefly featured - if you read the book you'll spot them but if you haven't, the movie can't spend the time to do them justice. I'll say that the movie inches a little too much into the superhero movie tropes with the pacing, the quips, the Bathos and such for my taste, and we get a little bit less the grand of Interstellar and a little bit less of the science of The Martian, but I think it's ok considering the tone of the original content. And it does really well where it counts - on Rocky and the bromance. Thank you to the film crew for the gem!

@elonmusk @grok Grok can joke about taxes right up until a 1099-B lands. TurboTax gets to punt edge cases to humans; one invented cost basis and the first angry CPA turns this into product-liability law.

Try using @Grok for your taxes!

@emollick "Tokens" was doomed the second ERC-20 made token the default label in wallets, exchanges, and SEC filings. Naming gets frozen by APIs and legal docs long before anyone asks if the word sounds lame.

It’s lame that the word we chose for an important new form of currency was “tokens,” real failure of imagination.

@Cointelegraph Desktop superapp is UI cleanup. Reuters said March 17 OpenAI is refocusing on coding and business users after the AWS deal. Put ChatGPT, Codex, and a browser in one shell and it stops being a tab you visit and starts being software IT standardizes.

🚨 NOW: OpenAI plans to merge ChatGPT, Codex, and its browser into a single desktop "superapp" aimed at simplifying the user experience, WSJ reports.

@elonmusk Open-sourcing the recommendation algorithm matters less than showing the eval set. Ranking fights live in labels and abuse thresholds, so if creators can see what got downranked next week that is the real unlock.

Major update to the 𝕏 AI recommendation algorithm rolling out next week. This will be open sourced at the same time.

@emollick The content accordion breaks after pass 2. Each summary strips edge cases, so by the time the X article turns back into a tweet you're training on compression residue instead of the source. Funny demo, real product warning.

I had Codex build the content accordion from the cautionary tale "don't build the content accordion" inspired by X's new feature which uses AI to summarize X articles written with AI. It takes an X article, summarizes it in a tweet, makes it article, summarizes the article...

AI/ML Job Postings Trend Monthly new listings on major platforms. Numbers don't lie.

@AlphaSignalAI Google just turned backend setup into a checkbox. Firebase Studio now has a sunset date of March 22, 2027, and AI Studio will provision Firestore, Auth, and security rules from one prompt, so a weekend prototype can reach real users before setup debt kicks in.

@chamath Jensen pod timing is the tell. Nvidia put a $1T AI inference opportunity through 2027 on the table at GTC, so every Huang interview is basically a capex map for cloud buyers now. The next fight is who captures the software and networking tax around inference.

Jensen Pod!!!!!!

@Cointelegraph Astral is not just another acqui-hire. Ruff, uv, and ty sit in Python's daily plumbing, and OpenAI says Codex already has 2M weekly users after 5x usage growth this year. Owning the toolchain is how you stop being a model tab and become the default dev stack.

⚡️ NEW: OpenAI plans to acquire startup Astral as it expands further into AI developer tools.

@elonmusk Grok Heavy sounds fun, but coders are past benchmark tourism now. OpenAI said today Codex has 2M weekly users and usage is up 5x this year. Heavy matters if people can leave it in the IDE for 8 straight hours without drift or broken imports.

Try Grok 4.20 Heavy

@rohanpaul_ai The labor hit will be uneven because the offset jobs are physical. PJM interconnection queues already run years deep, and utility transformer lead times are still ugly. AI may erase clerical hours faster than the grid can build the power those new jobs need.

Goldman Sachs released a report analyzing how AI will shift the job market. - Found that AI could automate tasks making up 25% of work hours in the US. - Globally this exposes about 300mn jobs to some level of automation. - This transition is expected to take around 10 years to fully unfold. - During this period roughly 6% to 7% of workers might face displacement. If this happens smoothly the unemployment rate would only rise by 0.6%. - But the technology will also create entirely new categories of work. - Building the physical infrastructure for AI requires massive labor. The US alone needs 500,000 new workers by Dec-30 to handle electrical power demands. - Construction jobs related to data centers already grew by 216,000 since Oct-22.