Загрузка видео...
Не удалось загрузить видео
Launching Vorflux : The autopilot for software engineering. I was prev co-founder / CTO of Rippling ($10B) and #1 coder in India. Vorflux is my high octane Ferrari. Every AI coding tool still makes you fly the plane. That's the copilot model: you stay in the seat, approving every... show more
2,216,985 просмотров • 3 месяцев назад •via X (Twitter)
Комментарии: 73
2/ The models woke up. The models got genuinely superhuman at code this year. GPT-5.6 Sol just topped the coding benchmarks; a different lab led the month before; no single model sweeps. And they outgrew your laptop. The cloud you were offered is blind: it writes code on some box but can't run your entire app + services + test it. So every branch comes back to localhost, and the loop has you again.
3/ The Great Flattening Code stopped being the hard part this year. Follow that all the way out and the whole company changes shape: planning, design, review, execution all collapse toward the system that runs the work. One person inside, or zero. Your job stops being the task and becomes codifying the judgment that does it.
4/ The harness That judgment needs somewhere to live. It lives in the harness: the one place your engineering principles run on every session, lossless and compounding. We train no models, so it stays neutral. The labs ship their opinion. Nobody ships yours.
5/ Profile the human time Writing code became a commodity. Bottlenecks moved. Every model release, we profile where human time still goes and throw tokens at it until it's gone. Human time is the bottleneck. Tokens are not.
6/ Backlogs die first It was a pile of work waiting behind one person. Six bottlenecks sit between a prompt and a merged PR, and once they fall, the backlog becomes a choice: stop saying no, build them all, find out which ones matter. Still hiring engineers before you've maxed out tokens? You're playing the 2022 game.
7/ From prompt to code. Three bottlenecks between your prompt and working code: The machine. A real app is a dozen services at once; a blind cloud runs a script, not that. Vorflux stands up a raw machine like a new engineer's laptop and brings every process up live. Code that never ran isn't done, it's a guess. incl. iPhone/android emulators. Planning. Most human time is spent here now. The cheapest place to fix a mistake. You get a plan worth arguing with, torn apart by a model from another lab. Orchestration. Turning a pile of sub plans into one code change, each in its own fresh context.
8/ From code to shipped, while you sleep. Three more between working code and a merge that goes live: Testing. If the branch comes back to your laptop to check, what was the cloud for? The whole app comes up, a browser gets driven through the real flows, and you clear testing by watching the recording. Review. A model grading its own work always says it's excellent. Real review comes from a model in a different lab, harder on the work. I read the story of the diff from my phone and approve. The merge. Master never stops moving; the queue keeps pace, resolves the conflicts, lands change after change. Nobody rebases. The person who asked is still asleep.
9/ We run it with you None of this ships as a box with a good-luck note taped to it. We profile where your engineering breaks, wire the harness to your stack, and run the first sessions with you. The models churn every few months; keeping current is our job, not yours. Already have a Codex plan? Bring it in. It runs across the whole pipeline, planning, testing, review, merge, not just the terminal.
Tokenmaxx, not peoplemaxx. Sign up here:
Heres our full manifesto:
@vorfluxai @Rippling Blown away by my experience. You guys have created something special. Can’t wait for everyone to experience this. Excited to be along for the ride!
@vorfluxai @Rippling thank you so much for the support! means a lot ♥️
@vorfluxai @Rippling So it’s another agent harness? Which claims to remove the human in the loop yet operates by publishing PRs for a human to merge? I’m completely missing what makes this unique to any other agent harness
@vorfluxai @Rippling Thanks for the dialogue. The bottleneck now is no longer coding but other parts that involve humans -- planning, testing, reviewing, orchestrating etc. Please take a look at this and lmk what you think:
@hexednobility @vorfluxai @Rippling Does Vorflux support end-to-end ML engineering workflows such as data preprocessing, model training, hyperparameter tuning, experiment tracking, inference APIs, and MLOps, or is its scope primarily application development?
@hexednobility @vorfluxai @Rippling It can definitely help with a lot of these. I use it for all my ops, heuristic building, deployments etc etc.
@hexednobility @vorfluxai @Rippling why only slack? why not others? Email, whatsapp etc..
@hexednobility @vorfluxai @Rippling Hmm we need to add more. What would you like to see here? Cc @nntnag17
@vorfluxai @Rippling When @myprasanna drops a 5 min product demo you better listen. Congrats on 15M seed! 🙌
@vorfluxai @Rippling thanks a lot brother!
@vorfluxai @Rippling btw for those in doubt about #1 coder part here is Prasanna’s 2008-09 resume
@vorfluxai @Rippling thank you brother!
@vorfluxai @Rippling "i dont even know what acp is... I still dont" 😅 This looks really cool Prasanna. And it supports Codex? 👀 Great ad. Would love to check it out.
@vorfluxai @Rippling yes codex cuts down costs a lot. give it a spin! :)
@vorfluxai @Rippling Vorflux has been the single most transformative product we’ve used at Fizz for interacting with agents. Not just for engineers but across the entire org. Really excited to see where it goes.
@vorfluxai @Rippling Was super exciting working with you folks!
@vorfluxai @Rippling Congrats on the launch! We @PowersetRes love the product
@vorfluxai @Rippling @PowersetRes You are the best!
@vorfluxai @Rippling Congrats - exciting to see a company take a bolder bet on the future of coding!
@vorfluxai @Rippling appreciate your guys support @BorisMPower -- codex partnership has been huge for us.
@vorfluxai @Rippling Hi @myprasanna The biggest engineering bottleneck today is to bring the agent close to human emotions. No matter what you do a human is never satisfied and will find fault in systems. Help me set it up and test so I can clearly solve somethinh complex and demanding. cheers
@vorfluxai @Rippling Congrats 🙌 welcome back to the arena! :)
@vorfluxai @Rippling thank you brother!
@vorfluxai @Rippling Congratulations Prasanna. Wishing you success 🤞
@vorfluxai @Rippling thank you mam!
@vorfluxai @Rippling This looks amazing. Would love to try out!
@vorfluxai @Rippling thank you!
hey man if you build cloud coding agents for vorflux you could be saving hundreds of thousands per year on sandbox cost by using box ( its 5x to 25x cheaper than other providers while being superior for building cloud coding agents, see
@vorfluxai @Rippling I run one of the largest open source projects in GitHub - lfnovo/open-notebook. Would love to try your app
@vorfluxai @Rippling Congrats, Prasanna!
@vorfluxai @Rippling Thanks bro
@vorfluxai @Rippling been using it for few months at supermemory and you guys really cooked🔥. Congrats on the launch, prasanna
@vorfluxai @Rippling congrats on the launch! the video is so cool!
@vorfluxai @Rippling thank you sir! give it a spin :)
@vorfluxai @Rippling Love how you showcased the product details into the demo and skipped the valley standard 3-5 second animation haha ! Need some of Vorflux meet-ups in Blr like the 0xppl days !
@vorfluxai @Rippling lol yes! :) we should have a blr meet up
@vorfluxai @Rippling Congrats! Love the launch video and you in it!
@vorfluxai @Rippling haha thank you bro! :)
@vorfluxai @Rippling Congrats Prasanna!!
@vorfluxai @Rippling thank you sir!
@vorfluxai @Rippling Boom!!
@vorfluxai @Rippling really appreciate your amplification sir! grateful!
@vorfluxai @Rippling all of us at @supermemory have been lucky to see vorflux evolve from the very early days and it is easily the closest thing to an actual autopilot we’ve tried so far. congrats @myprasanna and team!
@vorfluxai @Rippling @myprasanna it can run and tests - ios and macos apps also on cloud?
@vorfluxai @Rippling Yes yes -- we can. no one else does it.
@asxna @vorfluxai @Rippling Congrats on the launch!
@asxna @vorfluxai @Rippling Thank you Ishan! Appreciate your guys early support.
@vorfluxai @Rippling Love this. Some of the features you mentioned are like full fledged products in itself.
@vorfluxai @Rippling thanks a lot for your early support brother!
@vorfluxai @Rippling Onward and upward!🚀
@vorfluxai @Rippling My biggest bottleneck is debugging performance regressions across distributed services. Correlating logs, traces, metrics, and code changes still takes hours. Curious how Vorflux would tackle root-cause analysis in a production environment.
@vorfluxai @Rippling Absolutely! Will work great. Would love to show it to you!
@defikaran @vorfluxai @Rippling Congrats on the launch!
@vorfluxai @Rippling Congrats Prasanna!
@vorfluxai @Rippling @myprasanna does it work well with legacy codebases? curious how it handles older stuff.
@vorfluxai @Rippling works really well! way better than how humans can
@vorfluxai @Rippling @myprasanna glad you think so! what surprised you most?
@vorfluxai @Rippling Rooting for Vorflux, Prasanna!🚀
@vorfluxai @Rippling thanks a lot bro!
@vorfluxai @Rippling Hey @myprasanna congrats on the launch. Just curious if your're in search for strong engineers / researcher too niche and v diff to find. If yes, we can help. Already helped some YC startups hire inference eng and research scientists
@vorfluxai @Rippling Don’t want to jinx it, but this has billion-dollar-plus potential written all over it. Huge congratulations on the launch! Wishing you an incredible journey ahead.
@vorfluxai @Rippling Congrats on the launch @myprasanna !! Cloud agents with great verification objectives is the future of coding!
@vorfluxai @Rippling for sure. most agents today don't even have verification capability today since they dont run the app with all its services end to end. lots to improve.
![hockeystick moments are the biggest opportunity of your career. It's exciting and kinda scary. Everything changes! I wrote Scaling Fast: Software Engineering Through the Hockeystick as your guide. 16 years of startup experience condensed into 240 pages Scaling Fast is based on startup war stories, academic research papers, talking to my mentors, and reading industry insights. It shows you what it takes to survive and thrive through the hockeystick. You'll learn about scaling teams and scaling code, how they influence each other, and why none of it matters if the business is bad. Scaling Fast: Software Engineering Through the Hockeystick is organized in 3 sections: - Scaling the Business - Scaling the Team - Scaling the Tech The business part, that's to help you evaluate the companies you join, this isn't a business book. The team part, that's key to shaping your everyday. We talk about delegating decisions, empowering engineers, working smooth instead of fast, finishing things all the way to done, good code review culture, and shipping incrementally without risky big bang releases. The tech part, that's my favorite. We talk about good abstractions, architectural complexity, observing your systems break, making steady improvements to your code without huge refactoring sprints, why tests don't solve everything, what's even worth testing, how your team structure impacts what you can do with the code, and why solving today's problem is more important than building for an imagined future. None of it is about numerical scaling [name|]. That's easy in 2025. Computers are fast. The challenge is building complex systems that don't fit in any one person's brain without breaking the business. oftware engineering when your whole company changes every 6 months is the fun part of this gig. And that's what Scaling Fast is about. 👉](https://image.24vids.com/tw-1998089442180161774/amplify_video_thumb/1998074091224641536/img/t4efROK8ifK2d1dk.jpg)
