正在加载视频...

视频加载失败

Introducing Omnara: the open-source alternative to Claude Managed Agents. Omnara is an API for creating and interacting with production-grade agents. You build the agent. Omnara handles hosting, durable state, sandboxes, permissions, secrets, and the rest of the production infrastructure. Available now

93,539 次观看 • 2 个月前 •via X (Twitter)

56 条评论

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

Sign up and build your first agent at Or have your coding agent do it. Paste this into Claude Code, Codex, or whatever you use: Read this skill and use Omnara to build me an agent: First make sure I'm logged in to Omnara, then ask what I want to build and suggest a few example ideas.

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

Your agent works on whatever machines you give it: our hosted sandboxes, your own sandbox provider, or anything running the Omnara daemon (your laptop, a VM, a box in your VPC). Model Providers: Anthropic, OpenAI, Bedrock, OpenRouter. Or any other compatible endpoint. Open-source models welcome. Bring your own keys or use managed credits. Want all of it in your own environment? Self-host the whole thing:

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

A few agents we've built on Omnara: — scans X daily for buying signals and DMs us the leads — writes our product usage report every morning on a cron — sits in Slack as a teammate and picks up whatever we tag it on — reviews PRs against our codebase's conventions — handles customer support, under your brand Source for the first two:

Asher Dale 的头像
Asher Dale2 个月前

Next level stuff 🔥🔥🔥

Kartik 的头像
Kartik2 个月前

First stop along the way to a country of geniuses in a data center

Mehul Fanawala 的头像
Mehul Fanawala2 个月前

Congratulations on the launch Ishaan. 🥳 Going to explore this.

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

🥳

Nikhil Prabhu 的头像
Nikhil Prabhu2 个月前

lets go hyped

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

hyped! lets build!

Joshua wong 的头像
Joshua wong2 个月前

pointed my agent at this and it told me it finally found a manager worth working for

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

AGI??

Christian Sparks 的头像
Christian Sparks2 个月前

🚀🚀🚀

Christopher Jones 的头像
Christopher Jones2 个月前

🔥 🔥🔥

Elias Stråvik 的头像
Elias Stråvik2 个月前

@never_settles_ super cool - huge congrats on the launch! how should one think about omnara vs. vercel + eve agents? when should one pick one or the other?

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

hey thanks! good question. tldr: eve, like mastra, is a framework for building agents with code. you author one agent as a typescript codebase and ship it as a deployment. omnara is an api for running agents as a service: each agent is a durable object with endpoints for launching and steering it at runtime. wrote more here:

Elias Stråvik 的头像
Elias Stråvik2 个月前

@never_settles_ amazing! read up on that thread but still don’t feel like i quite get it. i think even more examples of omnara in use would help - do you have a page with case studies or something? would love to learn more!

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

that's fair. the way i'd think about it: a lot of companies are building internal agent platforms right now, stripe minions, ramp inspect, shopify aquifer. each one is a service with endpoints their teams call to launch and interact with agents. omnara is that, open source. eve, mastra, langchain are for writing an agent in code and deploying it statically afterwards. omnara is a hosted service that you can create and interact with agents on. a platform. the docs walk through the api with worked examples:

Elias Stråvik 的头像
Elias Stråvik2 个月前

@never_settles_ love it - will read up more in the docs as well, thanks a ton!

Vatsalpandya333 的头像
Vatsalpandya3332 个月前

Lfggg!!

Angel Onuoha 的头像
Angel Onuoha2 个月前

Great launch, lets go!

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

LG!!

Tom Greenwald 的头像
Tom Greenwald2 个月前

Let's go Omnara 🚀

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

LG 🚀

FlowOps Daily 的头像
FlowOps Daily2 个月前

open source shows up clearly in the Ishaan Sehgal Introducing Om part, and the presentation is clean.

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

thanks! have @Hinson_Kwong to thank

Anshul Jain 的头像
Anshul Jain2 个月前

Been using the new platform for 2 weeks and it's one of the best AI infrastructures I've seen. Just pointed Claude at the docs and building new agents is super easy :)

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

awesome thats the goal :)

Aoden Teo 的头像
Aoden Teo2 个月前

Awesome work. Beyond just being open source, I think not having your managed agents be linked to a model provider is the move.

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

agree! own your agents

Sadok 的头像
Sadok2 个月前

Building the agent is the easy part compared to the infra

Divam Gupta 的头像
Divam Gupta2 个月前

This is so cool!

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

hype 🙌

Saurabh Jambure 的头像
Saurabh Jambure2 个月前

All the best 🚀

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

🚀

Chu You Chia 的头像
Chu You Chia2 个月前

Been playing around with this, looks pretty cool. Gotta do some more extensive testing and report Looks like it will dramatically reduce deployment friction and infra complexity

Hinson Kwong 的头像
Hinson Kwong2 个月前

This is huge let’s go!

Rohan Mukherjee 的头像
Rohan Mukherjee2 个月前

congrats on the launch!

Sam Brickman 的头像
Sam Brickman2 个月前

🔥🔥🔥🔥🔥

Vik Makarskyy 的头像
Vik Makarskyy2 个月前

Incredible work, gold standard for managing intelligence.

Justin Schroeder 的头像
Justin Schroeder2 个月前

congrats!

Chris Ife 的头像
Chris Ife2 个月前

@ycombinator Congratulations on your launch 🚀

dnu 的头像
dnu2 个月前

Open sourcing the hosting layer is a generous choice. It gives teams a way to inspect how long-running agents are managed instead of trusting a black box.

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

big fans of open source!

Pierson Marks 的头像
Pierson Marks2 个月前

Congrats on the launch 🚀

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

thanks! 🚀

Ofek Shaked | AI Engineer 的头像
Ofek Shaked | AI Engineer2 个月前

Durable state and sandboxes are the actual product. Permissions and secret handling is where these platforms usually leak first.

Aaron Makelky 的头像
Aaron Makelky2 个月前

sick video edit!

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

thanks! @Hinson_Kwong

Samrith Shankar 的头像
Samrith Shankar2 个月前

this is cool - internally we do the same at @withmitra, where the team can build agents directly using the MCP and the API

Dhruv Patel 的头像
Dhruv Patel2 个月前

Amazing. Congrats on the launch 🚀

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

appreciate 🚀

Cryptpal Singh | 🍊 的头像
Cryptpal Singh | 🍊2 个月前

lfg!

Ishaan Sehgal 的头像
Ishaan Sehgal2 个月前

LFG!

Anders Lie 的头像
Anders Lie2 个月前

The time to own your agents is nigh 🤖

boussecsou.x 的头像
boussecsou.x2 个月前

I'm lovin'it

aditya 的头像
aditya2 个月前

lfgggggg

相关视频

In the future, you’ll be able to accomplish a goal by just giving Claude an outcome and a budget. That’s the direction Anthropic is building in with its new Managed Agents features, announced at this week’s Code with Claude developer event. The basic idea: Claude, wrapped in a computer in the cloud, that you can spin up, scale, and manage as needed. Anthropic is taking on the infrastructure that kills most agent products, and making sure that it scales to meet the needs of agents running 24/7. On this week’s AI & I from Every 📧, I talk with Angela Jiang (Angela Jiang), head of product for the Claude platform, and Katelyn Lesse (Katelyn Lesse), head of engineering for the Claude platform, about what Anthropic is building and what it takes to make agents reliable in production. We get into: - Why the "build a generic harness, hot-swap any model behind it" playbook is already outdated. Angela points to eval data on Memory where the same task across different harnesses performed drastically differently. - The infrastructure wall every team hits in production—and why Katelyn thinks “my sandbox died and took the agent with it” is the real reason internal agents don't ship. - Why Anthropic is so bullish on using file systems and skills within Claude, including Angela's argument that those early design choices can compound for years. This is a must-watch for anyone trying to take an agent past the demo and into production. Watch below! Timestamps: How the Claude platform evolved from API to agents: 00:01:48 The primitives that make up Claude Managed Agents: 00:04:09 Why the harness and the model are becoming a single unit: 00:10:37 The infrastructure wall that kills most agent projects in production: 00:18:49 Why team agents need a different shape than individual productivity tools: 00:24:49 How Anthropic's legal team uses an agent to review marketing copy: 00:26:36 Using multi-agent orchestration for advisor strategies, adversarial pairs, and swarms: 00:34:24 How to measure agent success with outcome and budget as the end state: 00:35:50 What the platform looks like a year from now, when Claude writes its own harness: 00:39:11

Dan Shipper

66,921 次观看 • 6 个月前