bills right! brace yourselves for the typescript takeover @aparnadhinak from @arizeai, @dan_goosewin from @Vapi_AI, @lhchavez from @Replit, @dexhorthy from @humanlayer_dev, @JavaSquip from @Netlify ....and many others are coming to SF on nov 6 tsconf.ai

📣 It’s a very special day here at Replit- we are proud to introduce Agent 3, our most autonomous Agent yet 🔥 What's new with Agent 3: ⚡️Tests & fixes the App it builds automatically using a browser- checking buttons, forms, links and APIs. ⏲️Runs for 200+ mins autonomously as it is building, testing and fixing your app- eliminating the need for manual oversight. ⚙️Builds Agents and automations to automate complex & repetitive tasks. You can now build Slack or Telegram bots or workflow automations integrated with Notion, Linear, Sharepoint, Dropbox & more in plain English. Agent 3 starts rolling out to all users today. More details on whats launching today below 👇

Luis Héctor Chávez breaks down how security scans in agents add immense value! 🚀 Adding security isn’t just a step – it’s a critical enabler of goals while safeguarding your data. Watch the session now at #TestMuConf!

🎤 Time for the Keynote Session at #TestMuConf 2025 with @lhchavez, CTO of @Replit! 🚀 Discover How Software Testing can Increase Agent Autonomy and why tests are more vital than ever in the era of AI Agents. Don’t miss! Join now 👉 bit.ly/4oGcxIo

Get ready to master the technicalities of using software testing to perfect #AIAgent behavior with our keynote speaker, @lhchavez, CTO, @Replit, at #TestMuConf 2025 🔗 bit.ly/TestMUConference2025 In his session, Luis will explore the crucial intersection of #AI, #testing, and #automation and how #TestingStrategies improve AI agent reliability, reduce errors, and enable intelligent decision-making. Register your spot now if you want to gain insights on multi-modal testing, AI-driven decision-making, and the importance of continuous testing in modern #DevOps pipelines. #SoftwareTesting #TestAutomation #QualityAssurance #Conference

Acaban de publicar mi entrevista en Startupeable, donde hablé sobre mi trayectoria, IA, Replit, outsourcing y más. Aquí te dejo los enlaces de: 🎧 Spotify: rebrand.ly/SSP233INV 📺 YouTube: rebrand.ly/SYT233INV 🎧 Apple Podcasts:rebrand.ly/SAPP233INV

1/ De un pueblito en México🇲🇽 a liderar una de las startups más TOP de AI en Silicon Valley. La increíble historia de @lhchavez, CTO de @Replit

Anyone known folks at Squarespace who can help here?

i accidentally shipped a 🦈

LLMs have transformed the software engineering workflow, but fall short in providing precise answers around mathematics and code. To address this, @masadfrost and @lhchavez developed code-exec, a stateless environment optimized for quick math tasks, and agent-env, which offers a stateful environment for broader programming operations. code-exec The first approach is a stateless container server: replit.com/@luisreplit/eval-… This approach works best for simple math evaluation using Python and OpenAI’s GPT model for calling functions. You can use any customized Docker container image to create your environment and deploy it using Replit Autoscale Deployments. Read the full docs in the replit-code-exec GitHub repository: github.com/replit/replit-cod… agent-eval The second prototype is more stateful for broader programming operations. It uses a full Repl as the sandbox, which gives the agent: - the ability to install any packages - the ability to read and write files - the ability to read and write arbitrary code - complete control over the execution environment You can read the complete documentation here: npmjs.com/package/@replit/ag… Try it out, and check out the blog post here for more information: blog.replit.com/ai-agents-co…

ando en cdmx esta semana. avisen pa' vernos.

my hobby: coming up with ridiculous internal codenames (this is called "HUGE Repls" in the codebase and some design docs)

lifetime achievement unlocked: finally reviewed a C program bedded inside a yaml file that gets compiled to eBPF at runtime. what a wonderful world we live in.

for this incident in particular, we've been able to identify several places where we can increase reliability going forward. this is also going to influence future architecture decisions so that we can be more resilient by design.

we had a pretty bad incident yesterday at @Replit which affected all of our free users for over an hour. even though we like to have fun with the meeting names, we treat postmortems very seriously since it's an opportunity to learn about the weak points in our system.

My Hobby: still coming up with names for postmortem meetings 🧵

this saturday on #ReplCon i'm going to be giving a talk about ✨the future✨ of what you can build in Replit 🔮. make sure to register on replit.com/replcon2022 . see you all there!

hi @uptimerobot! we're seeing a massive drop of requests towards @Replit . any chance we could DM?

started working on a control plane, and someone said that it needed a Slack emoji to avoid typing it out all the time. well, folks, this is why you shouldn't let most engineers near the design tools. i've created a monster.

The terminal bell echoes the impermanence of all codes. (modified opening line of the Tale of the Heike. said after witnessing an AI-generated snippet of code that could not be recovered)