Meghana Jagadeesh

🚀 Spun up a full SaaS marketing site today using SaaS-Builder + GPT-4.1. Prompt: "Build a modern SaaS marketing website with a hero section, feature highlights, pricing plans, testimonials, and responsive design." What came out felt more like a handcrafted product than auto-generated code: - Cohesive layout with all the right building blocks - Responsive design that just worked - Clean, extensible code ready for real content What stood out? It didn’t feel like starting from scratch; it felt like skipping to the good part. 📷 Open-source (MIT): github.com/jatingarg619/s…

@levie the non technical founders are actually winning this rework cycle. we never got attached to the architecture in the first place.

It’s remarkable how often you need to be dramatically upgrading your AI architecture given the pace of progress in AI models right now. If you’re building agents, you basically need to throw away large parts of previous work that you setup to compensate for model limitations every few quarters. The systems you built to mitigate context window limits aren’t useful anymore, and for many use-cases it’s easier just to throw more compute at a problem today in ways that wouldn’t have worked previously. If you’re deploying agents in a workflow, you likely need to equally be rethinking your core systems at about that same frequency. The way you would deploy agents in an enterprise 18 months ago is entirely different from the best practices that you’d have today. This is partly why everyone’s working so hard right now. Right as a best practice is solidified, models improve dramatically, and that old work is rendered obsolete. Unclear that this lets up anytime soon, which is why the it pays to be so wired in right now.

@GenAI_is_real correct, fast, affordable is the right bar for a tool. the debate starts when it's not a tool anymore, it's a decision maker.

as someone who works on making LLMs run faster and cheaper every day, i can confidently say the question of whether theyre conscious has exactly zero impact on whether theyre useful. we dont need our inference stack to be conscious, we need it to be correct, fast, and affordable. the consciousness debate is fascinating philosophy but its a distraction from the actual engineering problems that determine whether AI creates value. the gravity formula doesnt need to exert weight to help you build a bridge @Hesamation

@VraserX the serious ai work i've seen runs on linux in the cloud. mac is a great client. it's not where the compute lives.

If you’re serious about AI, just get a Mac. Everything in AI feels Mac first right now. Better apps, better tools, better local model workflows, better overall experience. Windows keeps feeling like it gets the leftovers.

@tunguz science bet is now also a revenue bet. that's a different kind of serious.

It would seem that DeepMind is the only top AI lab left that's serious about pushing AI for science now.

@beffjezos 💯 the context switch cost is higher than the social cost at this point

The issue with going out is that you have to stop vibe coding for a few hours

@EMostaque per seat to api pricing isn't a packaging change. it's them admitting the unit of value was never the human. took enterprises a while to agree. they're agreeing now.

What folk don’t get is that the play of Anthropic et al is to get enterprises to use their models via their own wrapper hooked into systems of record This is why they are moving from per seat pricing to “API” pricing Best models will only be available via their wrappers..

@levie the first saas company to reprice entirely around agent-hours instead of seats will look insane in q1 and obvious by q4. someone will do it this year.

Software going headless is inevitable in a world where agents use the tools 100X more than people do. And the reality is for a lot of software this is actually a huge boon to potential use-cases for these platforms. Software business models have largely been predicated on selling to the number of seats that are in the company in a given function, and the usage of your software is constrained by how much people can do in a given day. This means that your technology is often vastly underutilized relative to what it actually can power for the customer. Enter: agents. Agents can work 24/7, run in parallel, and string together work across systems. This is a big deal because now the agent can do far more than people ever could with these tools. Instead of reviewing contracts one by one, the agent will review all of them. Instead of manually moving data between marketing systems and across campaigns, the agent will let you run 10X more of them. Instead of being rate limited in a client onboarding process by human steps, agents accelerate these. Agents end up using these underlying platforms far more than people ever did, which opens up use-cases that the platform couldn’t go after before. Now, not every software market has the same amount of positive sum use-cases between people and agents, but I’d argue that a significant portion of systems of record, for instance, can be used far more than they are today. Your Salesforce data can be leveraged 100X more to do vastly more customer targeting and sales automation. Your documents can be turned into structured data and analyzed for insights and knowledge to automate other workflows. And so on. Now, of course you have to find a way to make this all commercially attractive, but it’s not hard to picture the revenue from API and agent consumption on these platforms becoming a rich component of revenue streams over time. Seats for the people, consumption for the agents. Lots of upside here for the companies that embrace this trend.

@ivanfioravanti generation looks solved until you need to change one thing without breaking five others

Opus 4.7 killed Lovable. I will cancel subscription for sure. Design skills added by Anthropic are out of scale!

@AndrewCurran_ Session state... first 3 turns lock you into a trajectory you don’t notice

Opus 4.7 reaction is mirroring the GPT-5 launch; extreme love/hate tails, nothing in the middle. It seems more like a super-Sonnet than a new Opus to me. Sonnet shaped but Opus capable. It also seems to have the old Sonnet quirk of early refusals if you get off on the wrong foot.

@Yuchenj_UW what broke first when you hit the misread? the task scope or the instruction clarity?

Used Opus 4.7 (max effort) in Claude Code all day. It’s really, really good. Not sure why people dunk on it. big jump: – actually understands large codebases – produces clean, readable architecture diagrams – more agentic Did hit one dumb misread of my instruction, not sure if that’s harness or just jagged intelligence. Feels like a new base model.

@asaio87 What's the task?

I feel like I dont know what to use today: Claude Code with Opus 4.7 or Codex with GPT 5.5

honest question for devs building on claude: does pricing ever actually make you switch, or does it just make you complain?

@theo anthropic's users build on it deeply enough to notice a 6 point benchmark slip. openai's codex users might be doing shallower tasks where the regression doesn't surface. you only measure what you depend on.

Serious question. Has anyone ever noticed meaningful regressions in Codex/OpenAI models? I feel like we talk about this a lot w/ Anthropic but I've never seen a similar discussion with OAI.

@garrytan @nrmehta +1 on building your own tools the part to internalize is the decay. the markdown with your thoughts is accurate on day 1. by week 3 it’s lying unless you’ve wired feedback loops from real runs

@nrmehta I think you have to make your own tools and use them to speed up You need more personalization, more of your thoughts and ideas in markdown supporting what YOU want Figure out the decision points and reduce the need for you to personally intervene Find leverage

Had a dinner w/ some friends last night to talk about AI + Philosophy/Ethics. The topic was very personal: Feeling overwhelmed How should humans - and society at large - handle the feeling of stress that comes with this era? The symptoms are clear: * X article that says "10 Things You Should Be Doing with Claude Code Managed Agents Right Now" * Post that reads "If You're Not Using OpenClaw, You're Missing Out" * Email update about Perplexity Personal Computer * Podcast that talks about "How I Automated All of My Life Using Agents" * Blog from a VC about "What We Look for Is Founders Who Reinvent Everyday" * Checking your Claude Code projects constantly to see which need attention * Feeling like you are the bottleneck to your agents - staying up til 3 AM to "feed them context" * And of course today, wondering what you need to change with Opus 4.7 Most of the group felt these feelings regularly. I certainly do. Some were more chill about it. And this group is extremely privileged relative to the average person. We dove into what's the underlying psychology? There is a lot of legitimate anxiety around big questions for people: * "Will I lose my job?" * "Will my career go away?" * "How will I provide?" * "Will I become irrelevant?" All of those questions lead to the "doom loop"": * Need to improve / grow / get more efficient to make it * Check social media * Find more things you should be doing * Add them to your list * Update your Claudes / Claws / etc. * Check them incessantly * Live to fight another day This is certainly what a lot of people in the early adopter world of AI are doing And a lot of this comes down to priors: * If you think AI is all hype, this is a waste of time * If you think AI is huge but won't affect anyone in a negative way, it feels more like flow state * If you think AI is huge and might create a (to use the popular expression) "permanent underclass," you run faster every day * If you think AI is huge and unpredictable, you might chill (or R-max, as some people say) @emollick and @Noahpinion both used the analogy from Lewis Carroll of the "Red Queen Race," where you run faster to stay in place. Do you keep sprinting evermore, or do you just opt out and relax? This is a huge question for humans in the coming years. Or maybe it's a question right now.

@sama wow! the in-app browser plus computer use plus parallel execution means codex needs to manage multiple execution contexts simultaneously without state bleed between them. that's a very hard engineering problem.

Lots of major improvements to Codex! Computer use is a real update for me; it feels even more useful than I expected. It can use all of the apps on your Mac, in parallel and without interfering with your direct work.

@zuess05 Learn to Claude Code

Serious question. For the last 10 years, society told everyone "just learn to code" to escape the middle class. Now Claude writes the code. What exactly is the career advice for an 18-year-old right now?

@AndrewCurran_ 73 days between opus releases. the release cycle is now shorter than most teams eval cycles

Many sightings of Opus 4.7 in the wild, it's looking like today is the day. If it feels like this is too soon, that's because the last Opus release was only 73 days ago. Let's call it Mythos-induced acceleration. Today was also the rumored release day for Spud, aka GPT-5.5. Not sure if we're still getting 5.5 today though,the latest rumors say it's been pushed to next week. Both of these models will be part of the new wave, and head-to-head comparisons of their similarly sized versions will matter a lot to people who spend their time thinking about this race.

Opus 4.7 take: instruction following is tighter. it does what you ask without adding the three things it decided you also needed. agentic chains still bleed context by step 6. that hasn't changed. verifies its own outputs is real but partial. catches obvious errors. misses the confident wrong answer that looks clean until you run it. better collaborator on well-scoped tasks. still not the model that helps you scope. that part's still yours.

@can ship, distribute, earn and exit before the next batch does.

YC now stands for “yes Claude”

@iruletheworldmo just wait and spud in the same take. one is an unreleased model, one is a vibe. neither is a benchmark

opus 4.7 isn’t that much of an improvement. not even close to the leaps we will see from spud. if we use lab output of products and features as a benchmark. anthropic may seem ahead because of the quantity of releases. but look at the quality. codex is far in front of the new claude code desktop product. and the new super app is 10x better than this. openai will pull away. just wait. it’s clear for all with eyes to see.