# Agentic AI Workshop Source: https://adityasinghal.com/agentic-ai-workshop-2025/ **2025 workshop archive** These materials describe the October 2025 course. 🚀 NYU CDS Community Workshop Master cutting-edge Model Context Protocol through live demonstrations. Build a personalized newspaper creation application using two collaborative agents. 3+1 Sessions + Showcase 90 Minutes Each 1 News Agent Project 📅 Wednesdays 5:00-6:30 PM 📍 Oct 1-22, 2025 🎬 Technical story-telling [View Highlights](https://www.youtube.com/playlist?list=PLgF7i4LH-YxYvhXK-yywN7eFRjRjQrq89) [View Public Repository](https://github.com/adityaarunsinghal/agentic-ai-workshop-2025) [Workshop Feedback](https://adityasinghal.com/agentic-ai-workshop-2025/feedback/) RECORDED IN CLASS ## Session recordings Watch the full session or find a topic in the chapter list. SESSION 1 October 1, 2025 ### Introduction to MCP & Tool Use [View on YouTube](https://www.youtube.com/watch?v=jNj2PI5A6zQ) Full session · 1:50:31 #### Find a moment Chapter links open YouTube at that moment. 42 chapters 1. [00:00 Welcome and introductions](https://www.youtube.com/watch?v=jNj2PI5A6zQ) 2. [12:15 Workshop overview and instructor background](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=735s) 3. [13:20 From "I have a Shih Tzu" (2021) to Agentic AI (2024)](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=800s) 4. [16:20 Participant survey results and expectations](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=980s) 5. [21:00 What is Agentic AI? The perception-action loop explained](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=1260s) 6. [23:50 ChatGPT breakthrough moment (Dec 2022)](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=1430s) 7. [24:30 GPT-4 with vision capabilities (Mar 2023)](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=1470s) 8. [25:00 Tool use emerges: Claude introduces structured tools (May 2024)](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=1500s) 9. [25:30 The integration nightmare before MCP](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=1530s) 10. [28:00 First tool demonstration: Getting current time](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=1680s) 11. [29:30 Why do we need schemas? Bridging LLMs and code](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=1770s) 12. [32:00 The breakthrough moment: MCP as universal standard](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=1920s) 13. [35:00 Live demo: Building a weather tool with MCP](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=2100s) 14. [38:00 Understanding MCP architecture through restaurants](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=2280s) 15. [43:00 Setting up MCP tools in VS Code](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=2580s) 16. [47:00 Brave Search integration demo](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=2820s) 17. [49:00 Real-time tool calling with Claude](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=2940s) 18. [50:30 Host (Application) = Restaurant Manager](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=3030s) 19. [51:00 LLM = Head Chef (makes decisions)](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=3060s) 20. [51:30 MCP Server = Kitchen (provides tools)](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=3090s) 21. [52:00 MCP Client = Waiter (facilitates communication)](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=3120s) 22. [53:30 Walking through the complete analogy](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=3210s) 23. [56:00 Project overview: Personalized newspaper agent](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=3360s) 24. [1:00:00 Use Case: "Fit or Flaw"](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=3600s) 25. [1:07:00 Project vision and requirements](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=4020s) 26. [1:07:30 Personalized news aggregation use case](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=4050s) 27. [1:09:00 Live coding: Creating MCP tools with FastMCP](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=4140s) 28. [1:14:00 The importance of discoverability in tool design](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=4440s) 29. [1:15:00 Testing the agent with real news sources](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=4500s) 30. [1:17:00 Tool composition: From 40+ calls to 5 calls](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=4620s) 31. [1:20:00 Homework assignment overview](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=4800s) 32. [1:21:30 Finding your own agentic use case](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=4890s) 33. [1:22:00 Building two MCP tools](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=4920s) 34. [1:23:00 Session 2 preview: Resources, prompts, and memory](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=4980s) 35. [1:24:00 Session 3 guest: Luca Chang (MCP core contributor)](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=5040s) 36. [1:25:00 Questions about tool execution and LLM behavior](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=5100s) 37. [1:28:00 Understanding the restaurant analogy deeper](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=5280s) 38. [1:30:00 Context window management strategies](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=5400s) 39. [1:32:00 Tools vs schemas: What the LLM actually sees](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=5520s) 40. [1:35:00 Handling tool failures and validation](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=5700s) 41. [1:38:00 Discovery and prompt engineering for tools](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=5880s) 42. [1:40:00 Multiple tool orchestration patterns](https://www.youtube.com/watch?v=jNj2PI5A6zQ&t=6000s) SESSION 2 October 8, 2025 ### Advanced MCP Features [View on YouTube](https://www.youtube.com/watch?v=TZ_UVMsj9Ls) Full session · 1:56:50 #### Find a moment Chapter links open YouTube at that moment. 26 chapters 1. [00:00 Session start and overview](https://www.youtube.com/watch?v=TZ_UVMsj9Ls) 2. [01:50 Student project discussions](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=110s) 3. [02:37 Library agent for multimedia archives](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=157s) 4. [05:00 Restaurant recommender with visual AI](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=300s) 5. [07:24 Fashion outfit agent with weather integration](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=444s) 6. [15:30 Enhanced newspaper agent overview](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=930s) 7. [21:40 Server instructions and context](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=1300s) 8. [28:00 Live demo: Creating personalized newspapers](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=1680s) 9. [40:00 Understanding model "laziness"](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=2400s) 10. [43:00 MCP Inspector walkthrough](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=2580s) 11. [48:30 Interest management system](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=2910s) 12. [52:30 Vector databases for article memory](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=3150s) 13. [53:30 Server-suggested workflows](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=3210s) 14. [55:00 Restaurant analogy: Weekend specials](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=3300s) 15. [56:00 Creating reusable templates](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=3360s) 16. [58:00 Restaurant analogy: Nutritional facts](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=3480s) 17. [1:01:30 VS Code integration demo](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=3690s) 18. [1:05:00 Preventing unwanted executions](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=3900s) 19. [1:07:00 Direct server-to-user communication](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=4020s) 20. [1:09:00 Live demo: User confirmations](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=4140s) 21. [1:12:00 MCP Sampling introduction](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=4320s) 22. [1:13:00 ChatGPT Apps SDK compatibility](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=4380s) 23. [1:14:00 HTML resources as interfaces](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=4440s) 24. [1:15:00 Smart tools architecture](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=4500s) 25. [1:16:00 Content ID efficiency pattern](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=4560s) 26. [1:19:00 Server-requested LLM reasoning](https://www.youtube.com/watch?v=TZ_UVMsj9Ls&t=4740s) SESSION 3 October 15, 2025 ### Multi-Agent Systems & Production Patterns [View on YouTube](https://www.youtube.com/watch?v=4YrRODYnYHY) Full session · 1:27:03 #### Find a moment Chapter links open YouTube at that moment. 45 chapters 1. [00:00 Welcome and attendance](https://www.youtube.com/watch?v=4YrRODYnYHY) 2. [01:50 Student project showcase begins](https://www.youtube.com/watch?v=4YrRODYnYHY&t=110s) 3. [04:00 Sports news agent with betting odds](https://www.youtube.com/watch?v=4YrRODYnYHY&t=240s) 4. [05:00 Restaurant recommender with visual AI](https://www.youtube.com/watch?v=4YrRODYnYHY&t=300s) 5. [06:00 Luca Chang introduction (MCP core contributor)](https://www.youtube.com/watch?v=4YrRODYnYHY&t=360s) 6. [08:00 Newspaper agent limitations review](https://www.youtube.com/watch?v=4YrRODYnYHY&t=480s) 7. [09:00 Token costs: Input vs output pricing](https://www.youtube.com/watch?v=4YrRODYnYHY&t=540s) 8. [10:00 Context length limits across models](https://www.youtube.com/watch?v=4YrRODYnYHY&t=600s) 9. [12:00 Context rot: Performance degradation with scale](https://www.youtube.com/watch?v=4YrRODYnYHY&t=720s) 10. [13:00 Needle in haystack problem: Tool selection](https://www.youtube.com/watch?v=4YrRODYnYHY&t=780s) 11. [16:00 Tool bloat: IBM research findings](https://www.youtube.com/watch?v=4YrRODYnYHY&t=960s) 12. [17:00 Real example: Claude tool confusion](https://www.youtube.com/watch?v=4YrRODYnYHY&t=1020s) 13. [19:00 Similar-sounding tools problem](https://www.youtube.com/watch?v=4YrRODYnYHY&t=1140s) 14. [23:00 Domain-specific agents concept](https://www.youtube.com/watch?v=4YrRODYnYHY&t=1380s) 15. [25:00 Extracting preference agent from news agent](https://www.youtube.com/watch?v=4YrRODYnYHY&t=1500s) 16. [27:00 Testing preference agent standalone](https://www.youtube.com/watch?v=4YrRODYnYHY&t=1620s) 17. [28:00 Add/remove interests implementation](https://www.youtube.com/watch?v=4YrRODYnYHY&t=1680s) 18. [32:00 Decoupling: No news mentions in preference tools](https://www.youtube.com/watch?v=4YrRODYnYHY&t=1920s) 19. [34:00 Vector DB for semantic preference search](https://www.youtube.com/watch?v=4YrRODYnYHY&t=2040s) 20. [37:00 Store preference with metadata (time, depth)](https://www.youtube.com/watch?v=4YrRODYnYHY&t=2220s) 21. [40:00 Semantic distance and relevance scoring](https://www.youtube.com/watch?v=4YrRODYnYHY&t=2400s) 22. [43:00 MCP client testing without LLM calls](https://www.youtube.com/watch?v=4YrRODYnYHY&t=2580s) 23. [45:00 Embedding functions and distance metrics](https://www.youtube.com/watch?v=4YrRODYnYHY&t=2700s) 24. [48:00 Reviewing agent specialization benefits](https://www.youtube.com/watch?v=4YrRODYnYHY&t=2880s) 25. [50:00 Wrapping preference agent as MCP server](https://www.youtube.com/watch?v=4YrRODYnYHY&t=3000s) 26. [51:00 Chat tool: Exposing agent capabilities](https://www.youtube.com/watch?v=4YrRODYnYHY&t=3060s) 27. [53:00 Producer-reviewer pattern explained](https://www.youtube.com/watch?v=4YrRODYnYHY&t=3180s) 28. [54:00 Content quality gates and validation](https://www.youtube.com/watch?v=4YrRODYnYHY&t=3240s) 29. [55:00 Reusability: Domain-agnostic design](https://www.youtube.com/watch?v=4YrRODYnYHY&t=3300s) 30. [56:00 Running news + preference agents together](https://www.youtube.com/watch?v=4YrRODYnYHY&t=3360s) 31. [1:00:00 News agent queries preferences at start](https://www.youtube.com/watch?v=4YrRODYnYHY&t=3600s) 32. [1:07:00 Preference agent performs semantic searches](https://www.youtube.com/watch?v=4YrRODYnYHY&t=4020s) 33. [1:10:00 Review process: Approve vs deny workflow](https://www.youtube.com/watch?v=4YrRODYnYHY&t=4200s) 34. [1:13:00 Haiku 4.5 vs 3.5 capability comparison](https://www.youtube.com/watch?v=4YrRODYnYHY&t=4380s) 35. [1:14:00 Preference agent denies initial report](https://www.youtube.com/watch?v=4YrRODYnYHY&t=4440s) 36. [1:15:00 News agent revision based on feedback](https://www.youtube.com/watch?v=4YrRODYnYHY&t=4500s) 37. [1:16:00 When to split agents: Three questions](https://www.youtube.com/watch?v=4YrRODYnYHY&t=4560s) 38. [1:17:00 Domain boundaries and security isolation](https://www.youtube.com/watch?v=4YrRODYnYHY&t=4620s) 39. [1:18:00 Complexity thresholds for separation](https://www.youtube.com/watch?v=4YrRODYnYHY&t=4680s) 40. [1:19:00 Parallel tool execution in agents](https://www.youtube.com/watch?v=4YrRODYnYHY&t=4740s) 41. [1:20:00 Multi-agent relationships and dominance](https://www.youtube.com/watch?v=4YrRODYnYHY&t=4800s) 42. [1:21:00 Crew AI vs Langchain comparison](https://www.youtube.com/watch?v=4YrRODYnYHY&t=4860s) 43. [1:23:00 Vector DB necessity for multi-agent](https://www.youtube.com/watch?v=4YrRODYnYHY&t=4980s) 44. [1:25:00 Context compaction strategies and quality](https://www.youtube.com/watch?v=4YrRODYnYHY&t=5100s) 45. [1:27:00 Claude vs Amazon Q compaction approaches](https://www.youtube.com/watch?v=4YrRODYnYHY&t=5220s) ## 🤖 What is Agentic AI? Agentic AI is all the buzz in 2025, both for businesses and employers. It refers to AI systems that can autonomously accomplish specific goals with limited supervision. These AI agents use tools, maintain memory, and mimic human decision-making to solve problems in real time, adapting their behavior based on context. This is a **"technical story-telling"** experience where you'll watch agents come to life through live demonstrations using pre-written Jupyter notebooks—then build your own newspaper agent as optional homework each week. ## 🍽️ The Restaurant Analogy 👨‍🍳 ### MCP Server = Kitchen All the tools and ingredients (resources) you need 🍽️ ### Client = Waiter Takes orders and delivers results (Claude, your app) 🧠 ### Agent = Chef's Expertise The intelligence that adapts and creates "A vending machine has fixed buttons. A restaurant adapts — same kitchen, infinite possibilities." ## Why This Workshop? 🧠 ### Live Demonstrations Watch real agents built through pre-written Jupyter notebooks that run with one click 🔧 ### Model Context Protocol Master MCP through FastAgent, Copilot, and Cursor integrations ⚡ ### Technical Story-telling Laptops closed - observe, discuss, and learn through guided demonstrations 🎯 ### Newspaper Agent Project Build a personalized news application with collaborative agents ## Workshop Journey 1 Oct 1 ### Building the Foundation From LLM Calls to MCP Tools Watch agents come to life through live demonstrations - LLM evolution & tool use fundamentals - MCP introduction & server setup - Building your first newspaper agent - Anti-patterns and teaching moments 2 Oct 8 ### Adding Memory & Context Vector Storage & Conversation Management Solve real context management challenges - Context bursting problems & solutions - ChromaDB for vector memory - MCP sampling and elicitation features - Conversation compaction strategies 3 Oct 15 ### Multi-Agent Collaboration Predictive Intelligence & Agent Orchestration Build collaborative newspaper creation system - Multi-agent newspaper personalization - Predictive intelligence patterns - Production hosting options survey - Agent collaboration workflows 4 Oct 22 ### Student Showcases Demo Your Creations Present your personalized agents - Student presentations - Peer feedback sessions - Advanced patterns & next steps - Production deployment strategies ## Tech Stack & Development Environment Pre-written Jupyter notebooks and modern MCP tools ### Tools & Platforms 🔧 FastMCP 🤖 FastAgent 💻 Copilot & Cursor 🧠 ChromaDB ⚡ Free OpenRouter Credits 📓 Jupyter Notebooks ### Key Learning Areas 🔧 Context bursting solutions 🧠 Conversation management ⚡ MCP sampling & elicitation 🏗️ Production hosting patterns ⚠️ Anti-patterns & debugging ## Workshop Structure 🎬 ### 15 Minutes Pre-Discussion Session kickoff with community discussion and context setting. 📚 ### 60 Minutes Live Demo Technical story-telling with live Jupyter notebook demonstrations. ❓ ### 15 Minutes Q&A Interactive discussion and personalized guidance for your projects. 🏠 ### Optional Homework Build your own newspaper agent using provided notebooks as foundation. ## Learning Outcomes This isn't just theory—you'll leave with: 🤖 ### Working Agents Functional agents you built yourself that solve real problems 🧠 ### Domain Knowledge Skills to build intelligent systems for any domain or use case ⚖️ ### Strategic Understanding Know when to use agents vs traditional applications 🚀 ### AI Revolution Position Be at the forefront of the next wave of AI innovation ## About the Instructors ### Adi Singhal Former CDS student and founder of the **NYU Data Science Club**. Currently at AWS helping build **Amazon Q**, the company's premiere agentic offering for developers and businesses. [Connect on LinkedIn](https://www.linkedin.com/in/adi-singhal/) ### Luca Chang Works in the **AWS Agentic AI organization** and is a regular contributor to the **MCP specification and SDKs**. Expert in building production-ready agentic systems. [Connect on LinkedIn](https://www.linkedin.com/in/luca-chang/) ## 🎯 What We Expect ✅ Attend sessions 💻 Engage with optional homework 🤝 Participate in discussions 🚀 Share your learnings A collaborative learning experience for builders interested in agentic AI. ## Ready to Explore Agentic AI? Join the CDS community in learning about agentic AI through hands-on demonstrations and collaborative exploration. [View Highlights](https://www.youtube.com/playlist?list=PLgF7i4LH-YxYvhXK-yywN7eFRjRjQrq89) ## Inspect the Hype ### Read the historical quotations and their sources - > In 2025, we may see the first AI agents 'join the workforce' and materially change the output of companies > > Sam Altman [Original source](https://forklog.com/en/sam-altman-openai-to-fully-roll-out-ai-agents-in-2025/) - > We're now confident we know how to build AGI as we have traditionally understood it > > Sam Altman [Original source](https://time.com/7205596/sam-altman-superintelligence-agi/) - > Discussion on ChatGPT and AI agents future > > Sam Altman [Original source](https://www.youtube.com/watch?v=nG-UnhwuIuI) - > Hey, why do I need Excel? I think the very notion that applications even exist, that's probably where they'll all collapse, right? In the Agent era > > Satya Nadella [Original source](https://www.windowscentral.com/microsoft/hey-why-do-i-need-excel-microsoft-ceo-satya-nadella-foresees-a-disruptive-agentic-ai-era-that-could-aggressively-collapse-saas-apps) - > With Copilot, we are bringing the power of next-generation AI to the tools millions of people use every day > > Satya Nadella [Original source](https://www.microsoft.com/insidetrack/blog/ai-powered-agents-in-action-how-were-embracing-this-new-agentic-moment-at-microsoft/) - > The age of AI agentics is here... a multitrillion‑dollar opportunity > > Jensen Huang [Original source](https://www.telecomtv.com/content/digital-platforms-services/nvidia-ceo-ushers-in-the-agentic-ai-era-52058/) - > AI agents are the new digital workforce > > Jensen Huang [Original source](https://blogs.nvidia.com/blog/ces-2025-jensen-huang/) - > GTC Paris 2025 content about agentic AI > > Jensen Huang [Original source](https://www.instagram.com/p/DLFOeTaNX3L/) - > Reduced support staff from 9,000 to 5,000; AI agents handle 50% of customer conversations > > Marc Benioff [Original source](https://finance.yahoo.com/news/salesforce-ceo-says-company-axed-180346796.html) - > Say hello to your new colleague > > Marc Benioff [Original source](https://www.linkedin.com/posts/marcbenioff_opinion-say-hello-to-your-new-colleague-activity-7313954417630662659-AiuP) - > Need fewer people doing some of the jobs that are being done today... agents are coming, and coming fast > > Andy Jassy [Original source](https://www.tekedia.com/amazons-jassy-envisions-agentic-future-says-ai-agents-will-soon-reduce-companys-corporate-jobs/) - > From a total reimagining of Search to Agent Mode, Veo 3 and more, Gemini season will be the most exciting era of AI yet > > Sundar Pichai [Original source](https://t.co/2n9rbGNj0Q) - > Disrupt your function with agentic AI by the end of the year. AI is the one thing that drives disruption > > Paul Hudson [Original source](https://www.mckinsey.com/industries/life-sciences/our-insights/dont-delegate-the-ai-revolution-a-conversation-with-sanofi-ceo-paul-hudson) - > The technology to build powerful agents is already here... Agentic AI is the catalyst that can make this transition possible > > Arthur Mensch [Original source](https://www.mckinsey.com/capabilities/quantumblack/our-insights/seizing-the-agentic-ai-advantage) - > Agents... will upend the software industry, bringing about the biggest revolution in computing since we went from typing commands to tapping on icons > > Bill Gates [Original source](https://interface.ai/) - > These more agentic systems are going to become possible, and it is why I think 2025 is gonna be the year that agentic systems finally hit the mainstream > > Kevin Weil [Original source](https://interface.ai/) - > When I hear '2025 is the year of agents,' I get nervous... this is the decade of agents > > Andrej Karpathy [Original source](https://www.accenture.com/us-en/blogs/data-ai/beyond-hype-why-agentic-ai-closer-than-you-think) - > 'Agentic' is the latest buzzword in AI, and in this case it's appropriate to believe the buzz > > Cole Stryker [Original source](https://www.ibm.com/think/insights/agentic-ai) - > Think of agents as the apps of the AI era... agents will begin to transform every business process > > Charles Lamanna [Original source](https://news.microsoft.com/) - > Agents are the new apps for an AI-powered world. Every organization will have a constellation of agents > > Jared Spataro [Original source](https://www.microsoft.com/insidetrack/blog/ai-powered-agents-in-action-how-were-embracing-this-new-agentic-moment-at-microsoft/) - > The next wave in the evolution of generative AI... able to tackle complex, multi‑step problems > > Justin Boitano [Original source](https://www.inkl.com/news/nvidia-is-jumping-on-the-agentic-ai-bandwagon-with-new-blueprints-to-simplify-work) - > Agentic AI is the next frontier of AI development... requires full-stack optimization > > Ahmad Al-Dahle [Original source](https://blogs.nvidia.com/blog/nemotron-model-families/) - > Unlock a whole new level of enterprise productivity beyond today's generative AI scenarios > > Philipp Herzig [Original source](https://blogs.nvidia.com/blog/nemotron-model-families/) - > AI agents make it possible for organizations to achieve more with less effort, setting new standards for business transformation > > Jeremy Barnes [Original source](https://blogs.nvidia.com/blog/nemotron-model-families/) - > Think about your org chart. Now imagine it features both... humans... and AI agents. That's not science fiction; it's happening > > Jorge Amar [Original source](https://www.mckinsey.com/) - > Half of the respondents say that more than 50% of AI deployment will be autonomous in their company in the next 24 months > > James Brundage [Original source](https://www.ey.com/) - > It's written by AI agents. It's mood-boarded by AI agents. The director of photography is an agent > > Wesley ter Haar [Original source](https://digiday.com/) - > In the long run, 'productivity is everything' and there is an 'insane need to go fast'. That is where agents come into play > > Arvind Krishna [Original source](https://timesofindia.indiatimes.com/) - > Seeing a shift of AI moving from just being about Assistants you chat with to Agents that go out and execute work for you > > Aaron Levie - > The agentic AI market explodes from $5.1B to $69B by 2032. That's 13x growth in 7 years > > Karl Mehta [Original source](https://t.co/eEXTR741Zq) - > Amazon Bedrock AgentCore announcement > > Swami Sivasubramanian - > How to Rewrite the Future and Win With Agentic AI > > Dan Gottlieb - > Agentic AI is here. But the real question isn't what to automate, it's which decisions to > > McKinsey [Original source](https://t.co/TzerjON53Q) - > Next phase of AI computing > > NVIDIA [Original source](https://x.com/nvidia/status/1932723231821017399) - > Agentic AI as #1 strategic technology trend for 2025; 80% automation of customer service by 2029 > > Gartner [Original source](https://www.pagerduty.com/resources/insights/analyst-report/gartner-report-top-strategic-technology-trends-for-2025-agentic-ai/) - > $2.6-4.4 trillion annual GDP contribution by 2030 > > McKinsey - > 25% of enterprises using GenAI will deploy AI agents by 2025, growing to 50% by 2027 > > Deloitte [Original source](https://www.deloitte.com/us/en/about/press-room/deloitte-technology-media-telecom-2025-predictions.html) - > 67% of C-level executives expect autonomous agents; 58% already integrated > > BCG [Original source](https://www.vamsitalkstech.com/agentic-ai/bcg-survey-agentic-ai-reshaping-it-spending-priorities-in-2025-key-insights/) - > $13.81 billion in 2025 to $140.80 billion by 2032 (39.3% CAGR) > > MarketsandMarkets [Original source](https://www.marketsandmarkets.com/Market-Reports/agentic-ai-market-208190735.html) - > $5.25 billion in 2024 to $199.05 billion by 2034 (43.84% CAGR) > > Precedence Research [Original source](https://www.precedenceresearch.com/agentic-ai-market) - > Global agentic AI market to reach $51.8 billion by 2028 > > IDC [Original source](https://my.idc.com/getdoc.jsp?containerId=prUS53765225)