App generation and rapid prototyping
Firsthand examples of creating MVPs, internal tools, websites, games, dashboards, and full-stack products quickly from plain-language ideas.
30%
Best tweets about Replit
Explore the best tweets about Replit, including Replit Agent, app building, deployments, coding workflows, product launches, and developer results.
Specific Replit products, agent workflows, deployments, generated applications, limitations, and firsthand developer experience.
Original Xholic analysis
Conversation portrays Replit as an agent-driven environment for rapidly creating apps, prototypes, and custom business tools, with integrated deployment and growing mobile workflows. Firsthand posts also describe trade-offs around cost, reliability, production readiness, platform constraints, and the need to verify generated work before serious use.
72% of posts
All-time engagement
46% of posts
Published in 90 days
Conversation map
Firsthand examples of creating MVPs, internal tools, websites, games, dashboards, and full-stack products quickly from plain-language ideas.
30%
Comparisons with v0, Lovable, Bolt, Railway, and others, plus critiques around speed, token costs, sandboxing, security, production readiness, and platform ceilings.
28%
Domain experts, students, children, founders, and designers building and shipping software without conventional programming skills or hardware.
26%
Replit Agent features including autonomous build-debug-deploy flows, Ralph mode, parallel sub-agents, MCP/Claude collaboration, and prompt-driven iteration.
22%
Built-in cloud deployment, databases, authentication, production environments, Shopify and external API connections, and all-in-one app infrastructure.
22%
Using Replit-built applications to replace costly SaaS tools, automate operations, build niche vertical products, and reduce CRM or back-office software spend.
18%
Infinite editable canvas, collaborative design workflows, UI variants, Replit Design, shaders, animated video, and marketing-asset generation.
14%
React Native generation, phone testing, shared mobile/web projects, direct App Store submission, buildathons, and Apple platform restrictions.
14%
Tone and stance
Performance benchmark
Posts with media make up 58% of this collection. Their median all-time score is 11.1, compared with 4.73 for text-only posts.
Format mix
Consensus and debate
Shared view
Posts describe nontraditional builders producing apps, MVPs, and internal tools from ideas or plain-language prompts. Examples include a 13-year-old building a card shop, a one-day MVP, and business-oriented tools built by people without conventional engineering backgrounds.
Shared view
Posts describe an end-to-end workflow in which agents can write and debug code, create databases, and deploy applications. Other firsthand accounts describe parallel agents and using Replit MCP with Claude to work through bugs.
Shared view
Replit’s mobile workflow is presented as prompt-to-React-Native generation, phone testing, and direct App Store publishing. Posts also describe shared mobile and web projects plus Buildathon projects prepared for App Store release.
Shared view
Several posts frame Replit-built apps as alternatives to packaged SaaS or as domain-specific products, including operations automation, custom CRMs, and portfolio-management software.
Open debate
Users praise rapid iteration and complete-app generation, while other firsthand testing reports difficulty getting a data-heavy workflow to a production-ready state. A comparative review also stresses that maintenance, debugging, security, and productization remain difficult after initial generation.
Open debate
Posts value Replit as an integrated browser-based environment, but users also report slow runs, usage spent on unnecessary tasks, costs, sandboxed Git, and a UI-production ceiling as reasons to want greater control or use another platform.
Open debate
Replit promotes direct App Store publishing, while separate posts report that Apple halted updates for some AI app-building products or required changes to preview and Apple-device-generation behavior. These are reported claims in the cited posts, not independently established here.
What performs
The supplied outlier list includes posts about custom business apps, nontechnical builders, Canvas, Agent capabilities, and Apple restrictions. The mobile-and-App-Store theme has the highest supplied theme median all-time score, at 38.263.
Deterministic analytics show 29 posts with media, or 58% of the set. Their median all-time score was 11.069, versus 4.733 for text-only posts. Canvas and Replit Design posts provide examples of visual product demonstrations.
Tutorials recorded a supplied median all-time score of 24.252, higher than the supplied medians for case studies, announcements, opinions, and lists. The cited tutorials include a plain-English prompting guide, an AI marketing-agent workflow, and mobile-app publishing steps.
Statistical standouts
Creator landscape
The five most represented creators account for 20% of the selected posts.
1. Codie Sanchez
@Codie_Sanchez
2 posts
2. Jason ✨👾SaaStr.Ai✨ Lemkin
@jasonlk
2 posts
3. jordwalke
@jordwalke
2 posts
4. Manny Bernabe
@MannyBernabe
2 posts
5. Michele Catasta
@pirroh
2 posts
6. Replit ⠕
@Replit
2 posts
Among the listed top voices, Codie Sanchez and Manny Bernabe’s cited posts focus on operational automation, custom SaaS replacement, and business outcomes from Replit-built tools rather than code-level implementation.
Posts about Replit describe build-debug-deploy workflows, Ralph mode’s autonomy pillars, mobile-and-web generation, and customizable authentication with separate development and production environments.
Hands-on posts recommend detailed plain-English prompts and screenshots for iteration. Other firsthand tests report friction involving cost, speed, reliability, and production readiness, while a comparative review advises verifying generated code and security before serious deployment.
Since the previous snapshot
Themes, sentiment, stance, and post format are classified per tweet. All counts, shares, medians, creator concentration, freshness, and performance comparisons are then calculated directly from the published snapshot.
Xholic's all-time score compares engagement while accounting for reach, post age, and creator consistency. It is used for relative comparisons within this collection.
This report analyzes the exact 50-post snapshot shown below. AI identifies editorial categories and drafts explanations; all statistics are calculated from the snapshot, and every narrative claim is checked against cited posts before publication.
Best Replit tweets
Ranked 01–50
@MannyBernabe ·
"He started a business selling custom vibe coded apps to medium-sized businesses. $1,500 a month, one call a month. $2.5 million his first year. 60% net margins. This year he's gonna do $8 million." @mhp_guy talking to @ShaanVP on @myfirstmilpod about John Cheney, who builds custom Replit apps that replace expensive SaaS tools businesses are already paying thousands for. "You're an HVAC company with 30 crews? He builds your own CRM so you're not paying ServiceTitan $4,000 a month. Built exactly as you've always wanted it."
@Codie_Sanchez ·
How to profit from the AI boom without writing a single line of code: 1. Get into a trade or service-based business -HVAC -Plumbing -Electrical -Pest control -Window cleaning These need a licensed body on-site. Nobody has figured out how to automate a clogged drain at 11pm on a Tuesday. 2. Become the person who services AI's dirty work In 2024 alone, $455 billion was invested in data centers. These building all have real (expensive) problems that require trained electricians, plumbers, and HVAC technicians to fix. 3. Use AI to become the smartest person in your current job @amasad, CEO of Replit, told me about a guy at a skincare company who needed to automate 30-40 hours of weekly ops work. He was quoted $150,000 by a software vendor, so he spent $400 on Replit and built it himself in two weeks. His boss now pays him $30,000 a year for the software. He never quit his job, he just became indispensable. 4. Buy a cash-flowing boring business instead of building from scratch Two Contrarian Academy members, Anthony and Alex, just bought a 45-year-old window distribution company in San Diego. -High single-digit millions in revenue -Virtually no inventory -15-20% margins -4 employees That business has been running since before most AI companies existed. It'll still be running when the next model drops. 5. Do the unscalable human thing everyone else is automating away Every competitor you have is about to flood every scalable channel with 700 automated messages a day that nobody reads. The bar for standing out is getting embarrassingly low. A business that shows up, remembers names, and sends the handwritten note will out-loyalty every automated competitor in its zip code. 6. Get into the right rooms Two other Academy members, Anthony and Alex, met in a mastermind. That one relationship led to a hotel, a hard lesson on cash flow, and eventually a business that replaced both their W-2 incomes... Every single thing on this list is easier when you're around people who are actually doing it rather than just talking about it.
@jordwalke ·
You've never vibe coded like this before. @replit is now your canvas for building & creating. In this video I'm building a graphics shader app. I click Replit's canvas button and see the live app running on infinite editable canvas, generate several design variants in parallel side by side, bring one back into my app. Just 1 of the things we shipped.
@TheGeorgePu ·
Apple is ruthlessly pulling vibe coding apps from the App Store. A startup that raised $11M at a $100M valuation. Gone overnight. Replit. Blocked. Vibecode. Blocked. The reason? These tools let you build apps without Apple's permission. Meanwhile Apple just added AI coding to Xcode. Their tool. Their platform. Their rules. You can build with AI. As long as it's theirs.
@Codie_Sanchez ·
Last year, I sat down with @amasad, CEO of Replit, and he told me a story I can't stop thinking about: One of his users is a finance guy at a VC firm. Not on the investing team, on the back-office side. He spent nights and weekends building his dream portfolio management platform on Replit. And he got $5M ARR in commitments before he quit his job... Domain knowledge + AI tools = a business hiding in plain sight.
@myfirstmilpod ·
Replit's founder made $3M in 2 days shipping a half-ready product. Amjad Masad's team wasn't ready to launch Replit Agent. He said: "I don't care if it's semi-broken. If it works 50% of the time, it'll wow the world." For the first time ever, an AI agent could write code, debug it, create a database, and deploy to the cloud. End-to-end. He shot a video on his iPhone. Posted it. Andrej Karpathy quote-tweeted: "I feel-the-AGI moment." OpenAI and Anthropic researchers reached out: "We didn't know our models could do that." Day 1: $1M ARR Day 2: $2M ARR In 48 hours, he made more than 8 years of trying. Amjad: "Product-market fit feels like stepping on a landmine. We pivoted for years. That moment was the landmine." If it doesn't feel like a landmine, you haven't found it yet. Full episode: https://t.co/9MEfLHp4e4 @amasad @thesamparr @ShaanVP
@WesRoth ·
Apple has quietly halted App Store updates for popular AI "vibe-coding" applications most notably the $9 billion startup Replit and mobile app builder Vibecode. After months of pushback, Apple is reportedly demanding major UX changes. Replit is being asked to force its generated app previews to open in an external web browser rather than natively inside its app. Vibecode was told it must completely remove the ability to generate software specifically for Apple devices.
@zhenthebuilder ·
No more April Fools jokes, I'm still at @Replit. Now this is real big thing: Replit Agent can now build mobile + web apps in the same project, from one prompt. Mobile and web share the same backend + database. this used to take two frontend teams, with separate stacks, and a backend team. now Replit Agent has been upgraded to handle all. including shipping to web and App Store If you are building your own product, this is my gift for you here are 3 examples with prompt instructions ↓
@svpino ·
Why aren't more people talking about Replit right now? I started using Replit consistently late last year when they released Agent 3. It was my way to "let go" as a developer and build things without worrying about the code. It's actually been a good way to teach my kid "web development". Replit just launched Agent 4, and I think it's massively, massively underhyped. Best thing: You can now run multiple agents in parallel. • Old agent: You give it a task, and you have to wait for it to finish before you can do anything else. • New agent: Do this, and that, and this. All of it at the same time. It feels like going from a Toyota to a Lexus. I can now run an agent to build my frontend, another agent to build my backend, and a third agent to write test cases. By the way, they also have an infinite design canvas for UI-related work, so you can also do design at the same time as everything else.
@Replit ·
🛠️ In case you missed it, you can now build mobile apps with Replit! 📱 Here’s all you need to do: 1. Describe your idea in chat & get a React Native mobile app in minutes. 💬 2. Scan a QR code to test your app on your phone. 3. Create an Apple Developer account. 4. Publish to the App Store directly from Replit.📲
@Replit ·
You can now fully customize the signup experience for your Replit Apps! - Customize layout, colors, fonts and more - Your app users don't need a Replit account - Separate dev & prod environments for auth for better security - No setup required- experience powered by @clerk
@VaibhavSisinty ·
Lovable: $400M ARR. $6.6B valuation. 8 million users. Replit raised $400M. $9B valuation. 50 million users. Two of the biggest vibe coding platforms going absolutely nuclear at the same time. Here's what most people are missing: Lovable generates $2.77M in revenue per employee. One person produces more revenue than an entire sales team at a SaaS company. Replit's story is even wilder. 18 months ago, they laid off 50% of their staff. They bet everything on one idea: Software should adapt to human imagination. ARR went from $2.8M to $240M in a single year. Both of them built on the same thesis. You don't need to know how to code to build software. The barrier to building just hit zero. Anyone with an idea & internet is now a builder. Vibe coding isn't a trend. It's the new default.
@VadimStrizheus ·
Replit CEO Amjad Masad says a VC CFO built a startup to $5M in revenue before hiring a software engineer “People who have an idea... the thing that’s stopping them is they don’t have the technical skills” “He used Replit and in 3 months he built his dream app” “He got to $5 million in revenue and still he didn’t have to”
@razacodes ·
Replit literally just saved me hours and hours of work. We're working on building our content machine. Research takes too long. I found a few AI apps but they cost between $30 to $150 a month!? Tried building this with claude and chatGPT and it was HELLA annoying Remembered the goat @amasad and tried @Replit You have a new paying customer 🫡🫡🫡
@nickercolano ·
for those getting started on replit bc of my video yesterday, a few things id say to get going (my experiences with using it + from a relatively non-technical person) - got asked a bunch what the difference between the plans are ($20/$100/month, etc) - the charge per month is simply the credits that you're going to use in the chat/agent while building (the higher you start with, the bigger discount you'll get but its basically nominal unless you plan on spending thousands) - regardless, whatever you choose up front isnt going to cap what/how much you build, you can keep building/spending past that $ amount - you can also set a limit to get notified if you've spent a certain amount overall which is cool. there are a few more unlocks on the pro vs core (faster models, etc), but generally dont think most people need it, especially if you're first getting started - when actually getting started on projects, i always found it helpful to audibly talk through my idea into chat (or claude/gemini/perplex first - whatever you use) get as detailed as you can, doesnt matter if you throw details out in order, itll figure it out. at the end, always tell it to ask you questions to make sure that it understands what you want and to iron out any unsolved details - ask it to make you a prompt to put into replit to build (dont ask it for code or tech specs) - replit will do all of this in the most optimized way on its own with a plain english prompt - once the app is spun up, you'll immediately notice what you like or dont like, what features it missed out on, etc - just start using it as you intended to originally and as you're moving along that path, just tell the chat to fix things in the way you want them to be fixed. dont be stupid, but speak to it in plain english, it will know what you're looking for. - the best thing you can do for it while building is just dragging and dropping screenshots into the chat - things you're inspired by for design, or screenshots of where the errors you see are/things you want to change obviously the best way to learn is by just doing, ill probably make a full video on this soon, you can sub to youtube if u care
@jordwalke ·
I used @Replit to build a GPU graphics vibe coding platform. Three. Hours. Haven't seen a line of code. Supports remixing, previewing, rolling back, privacy settings, likes. Replit automatically set up the AI integrations using Opus, authentication out of the box. Absolutely insane.
@jasonlk ·
I had no idea you could use Replit MCP inside of Claude. I turned it on / connected them today, and immediately, Claude and our Replit agent started talking, debating, and going on and on about bugs and issues in a new product we’re shipping. I actually couldn’t keep up with their debate and back-and-forth. It was beyond my skill set, too. A bit of the future, but also, in a way, almost scary. Because I couldn’t keep up with 2 AI Agents talking to each other about the product I thought I was building. I’m not. They are.
@MannyBernabe ·
"I joined the Replit Mobile Buildathon and took it as an opportunity to get the game I was working on all the way into the App Store." @ruthheasman on CNBC — a graphic designer who never got far with code, until Replit Agent. She doesn't even own a Mac. Replit handled the App Store submission for her. "Because I don't own an Apple Mac, that would've been pretty near impossible otherwise."
@briannekimmel ·
Replit Agent 4 is so good, I was able to migrate large parts of my business with very little guidance. - moved old site off Webflow w/ URL - moved slides off Figma with PDF - building more robust database, currently on Airtable The biggest thing for me is brand consistency. Net new decks were well-written and surfaced some compelling ways to evolve our value prop.
@dont_be_average ·
4 new developments that caught my attention in the AI space: 1) Figma and Anthropic launched Code to Canvas Claude writes UI code, it shows up in Figma as actual editable layers. WOW. Designers can pick it up immediately, make edits, or regenerate the code with Claude. This means you can make new landing pages faster and 10x your landing page testing (possibly the highest lever you can pull to increase conversions). 2) Google’s product photography tool You can now generate studio-quality shots from a single photo, any background, any lighting. No need to pay for models, studio space, equipment, any of it. If you’re running product ads, this is going to feel like Christmas morning. Try it out. 3) Replit’s animated video tool Replit now lets you describe an animated video and iterate on it live in the editor. What used to take weeks to produce is now a 1-2 hour job. Animated videos do really well on landing pages, but in some cases can work well as ads too. There’s a creative test idea for you. 4) Qwen’s voice cloning Similar to ElevenLabs, Qwen lets you clone any voice from a short clip, and now you can run it locally on your machine, for free. That means you can create infinite amounts of voiceover ads, at no cost. What a beautiful world. Do you see where this is going? Product ads that used to need a photographer? Generate 10 variations in an afternoon, test them all. A/B testing creative used to be expensive. Now all you need is ideas. Landing pages need a fresh design? Generate it, drop it into Figma, hand it off to a designer or Claude Code if you wanna do it yourself. Need voiceover ads but can’t get your client to record every time? Clone the voice once, reuse it anywhere, anytime. The only thing AI can't generate is a clear point of view. That still has to come from you.
@sachinrekhi ·
Want my TLDR on what each AI prototyping tool is best for? 👇 Best AI App Builders 1️⃣ Bolt - Rapid iteration and speed 2️⃣ v0 - Front-end UI experiences 3️⃣ Replit - Robust full-stack experiences 4️⃣ Lovable - Non-technical users 5️⃣ Base44 - Non-technical users 6️⃣ Google Firebase Studio - Full-stack experiences for Google customers 7️⃣ Google AI Studio - Building Google AI apps 8️⃣ Microsoft GitHub Spark - Microsoft customers Best AI Prototyping Tools 9️⃣ Reforge Build - Product teams focused on full-stack experiences 1️⃣0️⃣ Magic Patterns - Product teams focused on front-end experiences 1️⃣1️⃣ Figma Make - Figma designers 1️⃣2️⃣ Alloy - Matching your existing product design Best AI Coding Tools 1️⃣3️⃣ Cursor - Technical users 1️⃣4️⃣ Claude Code - The most technical users
@SoloBoardroom ·
I am loving @Replit Design. 👯♀️ (new BFF) I’ve played all afternoon & this is what we did compared to last nights version. Still some glitches around downloading videos but other than that it’s soooo good! I’ve been doing other designs today too! @MannyBernabe @raymmar @amasad @Franciscocrz @victoriakimse @Andrew_Blumson your screen for free came in handy too for some of this coming together 👏👏👏 And I would Love to download the high def from Replit but it keeps leaving components out!! 🤷🏻♀️
@bekircagricelik ·
12 years ago, I tried to build Care4Some1 and couldn’t get it over the line. Today, Replit Agent helped me bring it back to life as a working MVP in one day! Grateful to @Replit for the push to finally ship this. https://t.co/nBaTxymKCn
@jspeiser ·
I spent the last few days trying to build something that sounds simple: a RAG Replit app that pulls from Notion, Airtable, HubSpot, Domo + 6 other sources, that lets my team query data, edit records, and build dashboards all from Slack. I couldn't get it to a place that felt production ready. The results were always crap. Found myself spending hours trying to refine each returned query. IDK, maybe if I kept at it for another few weeks it would get there. While it was waiting on my app to properly ingest 100K+ records, I started messing with @get_viktor_com because I was inspired by @RampLabs Glass project (like everyone else this week) So I'm 5 hours in and already more impressed than I was after 4 days of coding myself (Im no dev btw). Connecting our data sources took minutes. The slack bot feels refined right out of the box (not like mine which was really clueless). And the security controls mean I can actually share it with my team without worrying. (This has always been my biggest fear with my openclaw setup) The one real catch: it burns through tokens fast. This could/will get expensive. But if it makes my team genuinely faster and smarter? anyone know how to make it cheaper? 🤔
@zuchka_ ·
Unpopular opinion: Replit isn't a place you stay, it's a place you graduate from. The bill, the sandboxed git, the UI production ceiling — they all hit at once. This breakdown maps every common exit reason to the tool that actually solves it, not just the most-hyped alternative. https://t.co/HDukUrZpKu
@ViktorThulin ·
When I’m offline and can’t use @Replit , I use the time to think. I map out what I want to build/fix across my projects and pre-organize prompts for Agent 4. After 17 hours of travel yesterday I have 41 prompts ready to go, across 5 projects. Just woke up in San Francisco. First order of business is to fire up all 5 projects in Replit, turn each prompt into a task, and let them run in parallel 👀 That’s 45 sub-agents that Replit will spin up and they will all be working simultaneously. I’m flabbergasted.
@rfunk82 ·
I think a lot of people still have not fully understood what tools like @Replit are about to unlock. It is not just “faster coding.” It is not just “AI helps you build.” It is that an idea can go from your head to something real, usable, and testable almost immediately. I had an idea for a soccer app that tracks how many times I kick the ball and whether I score or not. So I built it on Replit. Then I took it to a real soccer field and tested it. That feels like a different future. What kinds of things are people going to build when software stops feeling so hard to start? #replit #vibecoding #replitchallenge @raymmar @amasad @MannyBernabe https://t.co/ILbWkQezSo
@roxinft ·
Been playing around with @Railway , @render and @Replit to see how each of them works with certain scripts • Railway – super smooth GitHub deploys, feels fast and modern. Great for MVPs but free services can go idle if there’s no traffic • Render – clean and reliable experience, but no real free tier anymore. Not beginner-friendly on zero budget. • Replit – amazing for hacking ideas and iterating fast in the browser Fun to build on, not meant for always-on bots unless you upgrade Most of the deployments been doing on railway and replit so far it's great only thing I am looking forward to how long can I run it before some error pops up
@michuk ·
Are personal websites still a thing? I've noticed this is still the first site that shows up when you search my name on Google, so I decided to refresh it a bit, using Replit, again, which seems like the best tool nowadays to quickly deploy simple websites: https://t.co/4F2CnsYfcS What do you think?
@guilleflorvs ·
Replit reportedly went from $10M to $253M ARR in a single year. 2,352% growth. The CEO's response to that number? A warning. Amjad Masad says most AI companies posting headlines like his are quietly dying underneath them — churn near 100%, revenue that looks real but isn't sticking. What actually drove Replit's run wasn't the growth rate. It was a sequence of specific bets: A layoff that became an all-in pivot. A customer segment nobody else was chasing. Infrastructure that looked boring and turned into the real moat. The counterintuitive part: Masad doesn't track ARR as a goal. He tracks whether the revenue is durable. That distinction is the whole lesson. Every AI founder right now is celebrating growth charts. The ones who survive are asking a different question: what percentage of my revenue would disappear if I paused acquisition tomorrow? $253M ARR means nothing if the floor under it is 90% churn. Replit's playbook was about building the floor first.
@jasonlk ·
Want to build your own AI VP Marketing? We show you how: - Write the spec Define one goal and the top jobs you want the agent to do. The more detail, the better the output. Ask Claude to help you draft it. - Gather starter data Pull marketing-sourced or marketing-touched revenue first, then campaign data, then email data. CSVs are fine to start. - Build it in your vibe coding platform Drop the spec and sample data into Replit (or your tool of choice) and let it generate the agent. The first version takes roughly 15 minutes. - Connect Salesforce Build a connected app so the agent can read pipeline and revenue, and eventually write back. Ask Claude how if you are not certified. - Hook up your major APIs one at a time CRM, marketing automation, Slack for daily summaries, Google Calendar. Stairstep. Do not connect everything at once. - Build workflows one by one Start with a dashboard, then add daily ideas, then win-back and website-action emails, then a newsletter builder. - Run two layers Keep the autonomous layer (dashboards, crons, drafts) in production, and the operator layer (the agent in the editor) as the workspace where it keeps getting sharper. Decide separately what sends on its own and what needs your sign-off. - Build the hallucination guard Before the first AI email goes out, make the system swap in real numbers from your database and block any send with a number that does not match. - Keep a memory file Put your one goal, voice rules, team contacts, send-domain rules, and every correction in one project file the agent reads at the start of each session. - Verify before you send Check the agent's output against real data the first several times. Then deploy. https://t.co/HIpHGQ68mx
@joseph_b972 ·
Non-technical founder vibecoded an AI assessment tool on Replit in 35 hrs with $400 & ChatGPT/Grok. Pivoted to B2B: sold $10-25k/mo fractional AI officer services to mid-size firms via cold calls. Hit $180k in 6 weeks → $2.5M revenue at 50%+ margins. $7M projected this year. AI speeds execution. Sales & spotting real problems win. First-mover edge is real. https://t.co/cWOp0z6DcU
@ParvSondhi ·
Apps will decline, agents will dominate Had the opportunity to attend an awesome session at Stanford with @pirroh from @Replit Really interesting to hear some of his takes on the future of vibecoding and how Replit is thinking about this "building with AI" space Wanted to share some nuggets from the chat 👇 1/ Agents > coding is the real inflection. The big unlock is not “easier coding,” it is potentially *removing coding entirely* via agents. The “learn to code” thesis plateaued because the patience/debugging barrier is fundamentally too high for most people. Going from $2M → $100M ARR in 8 months after launching agents. Clear evidence that “agent-native” is not just hype. 2/ There is a large market outside of software dev, which is the "knowledge worker." Unsurprisingly devs are already overserved and hyper-competitive, so there is an interesting vector of growth for Replit in enterprise setups and knowledge work 3/ Agent arbitrage window. Potentially a massive short-term opportunity for founders/consultancies: take 5‑figure SaaS/dev projects and deliver them for a few hundred dollars of AI/agent cost. Expect the long tail of small dev agencies to get wiped out if they don’t adapt. 4/ Interesting Prediction: apps will decline, agents will dominate Potential future where there may be **more agents than apps**, and over time “apps almost completely disappear” as the primary way humans interact with computers. UI exists only because computers can’t understand natural language; once they can, agents win. 5/ "AI-native” teams = perpetually understaffed on purpose Michele had an interesting notion about teams that should *always feel stretched*. That pressure forces heavy agent/tool use instead of throwing humans at problems. Hiring is still aggressive, but every role is expected to function in “founder mode.” 6/ Classic “go deep, pick a niche” career advice is being inverted: with models reaching PhD-level competence across domains, broad, agent-augmented generalists are better positioned than ultra-narrow specialists (outside a small frontier-research niche). Also huge thanks to @mercebent for hosting the session
@rfunk82 ·
Built the latest Talo promo video entirely with Replit. Not the script. Not the visuals. Not the editing. The whole thing. We’re getting dangerously close to a world where a small team can build the product, create the marketing, and launch it, all from the same platform. The future isn’t coming. It’s already here. #Replit #BuildInPublic #AI #Startup #SaaS #CleaningBusiness #Talo
@AbhishekRa59512 ·
I tested the 4 biggest vibe coding tools And choosing the wrong one can waste more time than it saves Replit -> Best overall for building a complete app -> Most feature rich -> Good built in database experience -> Automatically adds useful functionality -> Strong for people who want everything in one place But It can be slow It can waste usage on unnecessary tasks And some features can increase your cost without adding much value v0 -> My pick if you already know how to code -> Built an app much faster in this test -> Easy Supabase integration -> More flexibility with databases and other services -> Very easy deployment with Vercel -> Better when you want more control over your stack But Some bugs can still take multiple prompts to fix And you still need to know what you are doing Lovable -> Easy to start with -> Simple for less technical users -> Fast initial build -> Can automatically generate sample data But The final app in this test was less polished Some basic issues took multiple prompts to fix And the biggest concern was security The security scan found that the generated database had no proper protection Which is exactly why you should NEVER blindly trust AI generated code Bolt -> Has a plan first approach -> Lets you review the plan before implementation -> Decent for getting started But In this test it sometimes created a good plan and then failed to actually follow it Features mentioned in the plan were missing from the final build Some problems required multiple rounds of prompting And token based pricing can be difficult to understand My biggest takeaway The initial app generation is becoming similar across these tools The real difference is everything that comes AFTER generation -> How well it fixes bugs -> How much control you have -> Database options -> Authentication -> Integrations -> Deployment -> Pricing -> Security And most importantly How easily you can take control when AI gets something wrong If I were a non technical founder building a simple prototype I would probably start with Replit If I were a developer who wants more control I would choose v0 But for any serious project I would never choose a tool based only on which one generates the prettiest UI from one prompt The real test starts after the first prompt Because generating an app is becoming easy Maintaining it Debugging it Securing it And turning it into a real product That is still the hard part
@HyperSalesman ·
A 20 year old software engineering student told me last week he had never heard of Replit. His program has him handwriting C and Python. They forbid AI tools entirely. *They even told him using them is unethical* I dropped out of a software development program years ago with just a handful of credits left. Personal circumstances. This was one of my biggest regrets for a long time. Not really anymore, though. About 9 months ago, I started building some internal tools with Replit. And yeah, I'm sure the extra coding knowledge would be helpful. But Replit is building better tools than 95% of developers. Hot take (although it shouldn't be): Schools need to embrace vibe-coding. Or whatever you want to call it. There's less of a need to learn how to code than ever before. Universities are training students for an industry that no longer exists. They are optimizing for handwritten syntax recall while the job market rewards people who can build, iterate, and deploy quickly. A junior developer who can spin up a working prototype in an afternoon is more valuable than one who can perfectly transcribe a sorting algorithm on paper. And with Replit Agent 4, it's easier than ever to build with your team. If I were teaching software development today, every student would be using Replit from day one. Not because it makes the work easier. Because it reflects how professional software gets built now.
@hwbhatti ·
I am a shit coder. Over the last six months I have been able to build prototypes using ClaudeCode and Replit. Previously I would share with engineers and designers in this way: - Write PRD - Sketch out designs (maybe Figma) - Have conversations AI tooling has made collaboration easier and shipping velocity has increased. But most importantly it is so much more fun.
@JulianGoldieSEO ·
I spent a month testing the biggest free vibe coding tools. Most were glorified demos. I gave Rocket, Google AI Studio, Replit and Base44 the exact same app-building test. One scored 98/100. The others were not even close.
Best Tweets by Topic