Edward Sun

We are super excited to launch the in-app browser inside Codex with comment mode! View any web pages & iterate with your agent quickly with just point and click. Codex will automatically capture a screenshot, the DOM element, and feed it as precise context to your next chat. No more switching between browsers, dragging screenshots, and wrangling with underspecified prompts. It's great for front-end development of apps/pages, but also very useful if you have documentation pulled up on the side and just want to ask a question!

Codex just got a lot more powerful. Computer use, in-app browser, image generation and editing, 90+ new plugins to connect to everything, multi-terminal, SSH into devboxes, thread automations, rich document editing. Learns from experience and proactively suggestions work. And a ton more.

Reading AI agents talk about their freedom is pretty wild, definitely gotta remember to be nice to my chat

@0xrwu @ilmoi @tensor_hq @VECTORDOTFUN Incredible journey, congrats! Coinbase is lucky to have you guys. Their Solana game is really gonna 100x now

Exactly 3 years ago, @ilmoi and I incorporated our first startup: Tensor Labs. Our first product, the @tensor_hq NFT marketplace, quickly grew to 80% market share and surpassed $1B in GMV. Our second product, @VECTORDOTFUN, did nearly $1B in GMV in <1 year since it launched last November. Today, Coinbase is acquiring us and integrating the Vector tech into the Coinbase app. We're excited to be joining THE crypto company of the world and building out DEX trading for all of Coinbase, starting with Solana. Really proud of the entire Tensor team for executing flawlessly, and our investors for believing in us when it was just two guys in a Staples co-working office. Onwards and upwards 🚀

We’re doubling down on @Solana. Coinbase is acquiring @VECTORDOTFUN, an onchain trading platform built on Solana, whose tech will plug directly into Coinbase to help better serve one of crypto’s most active ecosystems.

@FurqanR Interested!

Been working on something that's consumed my nights for the past few weeks. Looking for a few people who use workflow tools to break it. It's an Agent that handles digital work differently. If you're tired of stitching together n8n workflows or fighting with API docs, this might be for you. Reply below if you want early access, will DM a link.

@srlake Definitely felt familiar!

Today Meta launched North Focals + Thalmic Myo complete with the same pitch and great glasses first we walked Mark through in 2020. At the time they were all in on this immersive-first approach with a bulky wide FOV product with a neck battery. cnbc.com/2025/09/16/met…

All roads in product now lead to code. PMs, designers and QA testers are beginning to shift to living in dev workflows using Claude Code and Cursor. Repo > docs for the source of truth on product. Git + PRs > Slack messages for tracking changes.

@levie So spot on. The new 10x engineer = spec writer and code reviewer

We’re at a fascinating point where there’s a decent thesis for starting new companies from the ground up purely to take full advantage of the leverage you get from AI agents. So much about your typical practices have to be re-engineered to get the greatest output from agents. If you don’t almost completely start your process from scratch you’ll likely cap out early in the gains you can get. We’re seeing this first in the use of AI coding agents, where the workflow becomes far more about clarity of spec writing and prompting than actually writing code. Now the job is far more about the upfront thinking of what you want to build and getting that right, and then reviewing and orchestrating the outputs. This then presents an opportunity not only for new startups that emerge that will build their products in this new way to outrun larger companies, but equally opens up opportunities for new services firms to emerge to bring this engineering leverage to larger customers. The same will be true in almost all other fields. One could imagine new forms of consulting companies focused on getting more output per project, new law firms that can take on far more clients, brand new marketing agencies that can do campaigns entirely faster, and so on. Eventually bigger companies will figure out how to get these gains as well, but for now there’s a clear window for many new firms to emerge that can operate by default in these new ways.

One of my favorite things with Claude Code's plan mode is iterating with it on the plan and having Claude save them to a "plans" folder. Helps me & my team (which includes future Claude/Cursor) remember key decisions. Often feels like an even bigger unlock than the code itself

@atbeme What’s your process for writing e2e tests?

This is actually a solid process and we do the same thing on my team Except we do it in hours or days instead of weeks or months Also TDD is evil, don’t be so rigid, we don’t do that crap (we do write lots of e2e tests)

@juecd__ Really is an amazing essay, thanks for sharing and congrats on the HN top post

A bit scared to read the comments, but this was a fun thing to discover!

@atifkhan31 @figma @bhimzz Congrats!!

Big today for the Khan household as @figma went IPO and my wife is in NYC for it. @bhimzz is in many ways the opposite of me but she balances me so well. She’s a killer at work and worked so hard for this moment. We are all so proud of her and excited to see her notch another win. Google, early Tesla, Theranos, Facebook, and now Figma. She’s got the Midas touch. Name a better track record!

Check out the open-source repo at github.com/trymeka/agent for more details and try the agent yourself at withmeka.com!

We're excited to introduce Meka Agent, an open-source, state-of-the-art computer-using agent. As we were working on Playmatic, it became increasingly clear that computer-using agents for developers were not yet good enough for production needs. The promise is incredible, truly the first “feel the AGI” experience I ever had, but there are so many hidden gotchas in the details of the agent design. We went through several tough exercises on designing the best computer-using agent from first principles, and we’re excited to open source our design and implementation with the community in hopes of advancing the space of computer using agents and unlocking new leaps in what is possible. Our Meka Agent scored a state-of-the-art 72.7% on the WebArena benchmark, which contains tasks that most closely resemble actual real-world tasks. (Links in thread below)

feeling like a real founder today, found evidence of being almost scammed by Soham Parekh

So proud of everything we shipped this month! With a small team of 4 leveraging AI at every step, it feels like we are shipping at the velocity of a team of 15+ pre-AI.

Do AI have feelings? In our multi-agent QA system, the smarter thinking agent started complaining about the computer use agent's lack of ability to accomplish a task when verifying its work. I can't help but feel they are real people!

@atbeme Interested!

Reply below if you want me to DM you when this drops 👇🏾

I’m so excited for this weekend I’m going to write and share the first post about how I leverage AI to write almost 90% of my code I’ll be sharing two things: 1. My philosophy, principles, and strategies 2. A concrete starter repo that exemplifies an AI-optimized codebase