CrewAI

CrewAI is #4 on the 2026 Enterprise Tech 30 Early Stage list. 🎉 The ET30 is voted on by 90+ of the top VCs and corporate development leaders in enterprise tech, so this recognition means a lot. But the real story is the same one it's always been: enterprise adoption of AI agents is accelerating fast, and we're grateful to be building in the middle of it. To our team: this one's yours. To our customers: thank you for trusting us with work that matters. To our partners: let's keep going. Thank you Wing Venture Capital and Eric Newcomer. Congrats to all the other listees! wing.vc/et30 #ET30 #AgenticAI #CrewAI

🔴 We're live — join the webinar now. The Latest with CrewAI Open Source is starting NOW. Lorenze Jay Hernandez (Lead OSS Engineer at CrewAI) is covering recent open source updates, what's shaping the next generation of agent development, and a forward look at agent harnesses and ecosystem trends. 👉 https://lnkd.in/gnaxWQjy #CrewAI #AgenticAI #OpenSource

This is my biggest one yet 👀 🫣 I've been going to ODSC in Boston for the past couple of years, following some of the most innovative minds in data science and AI. This year I'm honored to be on the other side as a speaker at ODSC AI East on April 29th at 11AM. If you're not familiar, ODSC is the #1 AI builders conference. Attended by practitioners, researchers, and engineers who are actually building, not just talking about it. I'll be covering: → Multi-agent architectures that actually work in enterprises today → How to decompose real business use cases into agent workflows → Scaling those workflows from pilot to production I'll also featuring something new from the CrewAI team!! If you're at ODSC East next week, come say hi. Would love to connect in person. Open Data Science Conference (ODSC) Sheamus McGovern CrewAI

We just shipped CrewAI coding agent skills. If you're building agentic systems with a coding agent today, you've probably hit the same walls: • Unclear starting points • Too many design decisions upfront • Inconsistent results when defining agents, tasks, and workflows Instead of making developers stitch together docs, patterns, and setup steps on their own, we packaged the guidance directly into the workflow. What's skills do you get? 1. 𝗴𝗲𝘁𝘁𝗶𝗻𝗴-𝘀𝘁𝗮𝗿𝘁𝗲𝗱: a clean on-ramp so your agent knows where to begin 2. 𝗱𝗲𝘀𝗶𝗴𝗻-𝗮𝗴𝗲𝗻𝘁: opinionated defaults for building focused, reliable agents 3. 𝗱𝗲𝘀𝗶𝗴𝗻-𝘁𝗮𝘀𝗸: guidance for scoping tasks that produce consistent results 4. 𝗮𝘀𝗸-𝗱𝗼𝗰𝘀: live access to CrewAI docs so your agent always has current context Works with Claude Code, Cursor, Codex, Gemini CLI, or any coding agent. Full guide: https://lnkd.in/gPbevV7m

Tomorrow we're going live with a breakdown of CrewAI's latest open source releases — the thinking behind them, and what they signal about where agent development is headed. Lorenze Jay Hernandez, Lead OSS Engineer at CrewAI, will walk through: → What's new across the CrewAI open source ecosystem → The reasoning behind our latest product decisions → How agent harnesses are shaping the next generation of AI systems → Our POV on building reliable, production-ready agents at scale 🗓 Thursday, April 23 | 10–11am PT 🎟 Register: https://lnkd.in/gnaxWQjy #CrewAI #AgenticAI #OpenSource

Everything your agent needs to ship a production crew. 𝗚𝗲𝘁 𝘀𝘁𝗮𝗿𝘁𝗲𝗱 𝗵𝗲𝗿𝗲: https://skills.crewai.com/

We just shipped CrewAI coding agent skills. If you're building agentic systems with a coding agent today, you've probably hit the same walls: • Unclear starting points • Too many design decisions upfront • Inconsistent results when defining agents, tasks, and workflows Instead of making developers stitch together docs, patterns, and setup steps on their own, we packaged the guidance directly into the workflow. What's skills do you get? 1. 𝗴𝗲𝘁𝘁𝗶𝗻𝗴-𝘀𝘁𝗮𝗿𝘁𝗲𝗱: a clean on-ramp so your agent knows where to begin 2. 𝗱𝗲𝘀𝗶𝗴𝗻-𝗮𝗴𝗲𝗻𝘁: opinionated defaults for building focused, reliable agents 3. 𝗱𝗲𝘀𝗶𝗴𝗻-𝘁𝗮𝘀𝗸: guidance for scoping tasks that produce consistent results 4. 𝗮𝘀𝗸-𝗱𝗼𝗰𝘀: live access to CrewAI docs so your agent always has current context Works with Claude Code, Cursor, Codex, Gemini CLI, or any coding agent. Full guide: https://lnkd.in/gPbevV7m

Building multi-agent systems is the easy part. Governing them at scale is where most teams hit a wall. Tomorrow I'm joining Galileo's co-founder Yash Sheth to talk about what happens after you ship your agents. The stuff nobody wants to think about until it's a problem: enforcing safety policies at runtime, controlling costs across agent fleets, keeping compliance teams confident, and managing behavior when agents hand off to other agents. We'll get into how to steer agents to the right models in real time, how to bring non-technical stakeholders into the policy loop without code, and how to govern first-party and third-party agents from one place. Last chance to register here: https://lnkd.in/gtfgub9J

Tomorrow: live webinar on governing multi-agent systems at scale. Building agents is one thing. Keeping them safe, compliant, and in control across your entire org? That's where most teams hit a wall. João (Joe) Moura (Founder & CEO, CrewAI) and Yash Sheth (Co-founder & CTO, Galileo) are breaking it down live — covering everything from runtime policy enforcement to bringing non-technical stakeholders into the loop. What's on the agenda: → Enforcing safety and security policies in CrewAI agents → Steering agents to the best models and fallback tools at runtime → Governing all your agents with one centralized policy layer → No-code policy management for risk and compliance teams 🗓 Tuesday, April 21 | 10am PT 🎟 Register: https://lnkd.in/gRjQNnwM #AgenticAI #CrewAI #AI

Have you noticed what AI is doing to user interfaces? Bear with me for a second... Software engineers have a long standing enemy called context switching. Your PM sends a DM asking you to fix a simple UI bug. You put it off for later. But you already context switched. Checking Slack opened a loop your brain refuses to close. This has a name: the Zeigarnik Effect. The fix is closing the loop faster, and AI is enabling that for the first time. At CrewAI, we provisioned an AI agent on Slack with access to our internal tooling. We just tag @Iris (sorry, Siri) and she can call crews, Linear, Claude Code, GitHub, send emails, etc. We're not the first ones doing this. Linear has a bot on Slack you can talk to. Salesforce is doing the same. Something is happening to user interfaces, and this is happening fast. The team already lives on Slack. Now the tools do too. No new interface to learn, no new tab to open, no branch needed. Just the thread you were already in. We keep asking what AI will automate. Better question: what will it make invisible? My take is that the interface itself is getting compressed. --- Working in AI means being wrong a lot. About what will work, what won't, what the field even is. I've started writing down the surprises as they come. Follow along if you like. 👋