CrewAI

The agentic future isn't coming. It's already here. We're powering over 60,000,000 AI agents every month across 150+ countries. I've spent the past year watching companies transform from curious adopters to what I call "agentic native" organizations. Early in the AI agent revolution, companies focused on technical questions - memory systems, tool calling, frameworks, architectures. But the conversation has shifted dramatically. Smart enterprises now realize they'll soon have hundreds or thousands of agents running in production. They're thinking about AI agents as organizational resources that need proper management – not just cool tech demos. This requires a complete enterprise suite for controlling these agentic resources. What does this suite need to address? Three critical factors: 1. Interoperability: No vendor lock-in across the stack 2. Observability: Complete visibility into agent actions 3. Governance: Strict controls over data access Unlike early LLM adoption (which happened at the edges of organizations), agents are being deployed centrally. Engineering teams establish control towers that enable different departments while maintaining governance. The results are staggering. Some organizations report 80-90% efficiency gains in key processes. That's happening right now, in production environments. This is why companies evaluate use cases based on direct ROI and cost savings. We're seeing thousands of use cases across every vertical – financial services, healthcare, media, government. What's interesting is how these use cases vary along two axes: • Complexity (simple vs. complex tasks) • Precision (creative freedom vs. exact requirements) This week, we announced our partnership with AWS - CrewAI agents can now natively call Bedrock Agents. The timing couldn't be better. In challenging market conditions, companies need efficiency more than ever. Open source solutions allow them to try before they buy. The winners won't be those with the most sophisticated AI. The winners will be those who effectively orchestrate agentic resources at scale. That's why we're building more than just a framework – we're creating the infrastructure for the agentic future. A future that's already arrived.

Where is your company on its AI Agent journey? Here's a slide I just shared in a webinar. :-)

Bring your laptops and join Lorenze Jay tomorrow to learn how to build CrewAI Flows!

🚀 Deploying your Crew to CrewAI Enterprise using the CLI 🧑💻 In this follow-up to our open-source getting started guide, Tony Kipkemboi shows you how to take your locally built Crew and bring it to production—with observability, metrics, and real-time control built in. 🎥 What you’ll learn: ✅ How to deploy your Crew via the #CLI ✅ Monitor every task, agent, and tool from a visual dashboard ✅ Use the AI Chat Builder to generate new crews with just natural language ✅ Track costs, token usage, and execution timing ✅ Trigger your crew from the UI 💡 Whether you're a solo dev or shipping production apps for enterprises—CrewAI Enterprise gives you the infrastructure to build, manage, and scale agentic workflows. 🔗 [Links in comments] #AIAgents #LLMops #Observability

🔍 New multi-agent build from the CrewAI community: a hotel search assistant that mimics human browsing — and it runs 100% locally. Akshay Pachaar just dropped a brilliant demo powered by: 🧠 CrewAI 🛜 Browserbase (headless browser automation) 🧱 DeepSeek-R1 via Ollama 💡 What it does: ‣ Parses queries like "hotels in NYC next weekend" ‣ Uses a headless browser to search #Kayak ‣ Pulls top 5 hotels + digs into pricing & amenities ‣ Summarizes everything in natural language 📦 Agents + tools: ‣ Hotel Search Agent (uses Browserbase to scrape live results) ‣ Summarization Agent (makes the data readable) ‣ KAYAK Tool (builds the URL from user input) Why it’s awesome: 🧩 No APIs needed — real-world scraping handled like a human would ⚡️ All local — no cloud LLM required 📈 Easy to customize and extend Full demo and GitHub in the comments — worth a watch if you’re building multi-agent research workflows, search assistants, or web-integrated agents. Great work, Akshay 👏

🎬 New Tutorial Just Dropped: Get Started with CrewAI Open Source! 🔥 If you’ve been meaning to try CrewAI but weren’t sure where to begin, this one’s for you. In this walkthrough, Tony Kipkemboi shows you how to spin up your first agentic app using #crewai open source framework. What you’ll learn: ✅ Installing CrewAI with lightning-fast "UV" by Astral ✅ Creating your first multi-agent crew with "crewai create crew <crew-name>" ✅ Defining agents, tasks, and tools with YAML ✅ Running the full workflow from the terminal ✅ Producing structured outputs like "report[dot]md" You’ll go from zero to your first working crew in under 15 minutes. 🎥 Watch the full video: [link in first comment] #OpenSource #AIagents #Python

🧠 Build full-stack AI agent apps with CopilotKit + CrewAI! 🔥 Developer’s Digest just dropped an excellent walkthrough video on how to connect the dots between CopilotKit🪁 and #CrewAI to build real-time, agent-powered UIs. 💥 The focus? Shared State – a pattern where your frontend and AI backend talk to the same source of truth. Modify something in the UI, and the agent knows. Agent replies? The UI updates instantly. Here’s what it covers: ✅ Build with Next.js + FastAPI ✅ Power the backend with CrewAI Flows ✅ Connect the UI using CopilotKit’s hooks and components ✅ Stream real-time LLM responses + structured state updates ✅ Learn how to cleanly separate schema between frontend & backend The video walks through a live demo turning a recipe assistant into a workout planner — showing exactly how to adapt schemas, function calls, UI labels, and more. This is a great one to learn from if you’re building: ✅ Embedded copilots in your SaaS app ✅ Frontend ↔️ backend agents ✅ Fully reactive UI driven by LLMs 🎥 Full tutorial + repo in the comments. #FullStackAI #AIagents #AIUX

📣 Multi-Agent Brand Monitor Agents(Runs 100% Locally!) 🧠 Akshay Pachaar is back with another banger—this time showing how to build a multi-agent brand monitoring system using CrewAI Flows + #DeepSeek-R1. The goal? Scrape web mentions across platforms like X, Instagram, YouTube, and websites—and generate platform-specific insights, all stitched together into a final report. 🧱 Tech Stack • ⚙️ CrewAI Flows for agent orchestration • 🌐 Bright Data to scrape at scale (no IP blocks, no captchas) • 🧠 DeepSeek-R1 served locally via Ollama 🧩 How it works • Scrape brand mentions using Bright Data • Route data to the right agents (platform-specific analysis) • Merge all insights into one powerful report 🚀 The demo shows just how powerful local, multi-agent workflows can be for real-time brand intelligence. GitHub repo link + walkthrough in the comments 👇 Nice work Akshay—super clean build! 👌 #crewai #opensource #ai

🔧 Build Your Own AI Agent Workflow with CrewAI + Streamlit 🎈 Brandon Krakowsky has put together an insightful walkthrough on creating a multi-agent AI workflow using CrewAI and Streamlit—all within a single visual interface. Check out the demo link in the original post below.

What an awesome week! 🚀 Thank you all who attended and tuned in to our first AI Agent Week Summit.