FK

Yeah! You can use @langchain with @golang After months of going through scarce materials and building lots of demos, I've been able to complete this LangChain course for Gophers. udemy.com/course/langchaingo…

Best Data book I ever bought In all my work with AI, one fact keeps standing out: “Data is King” And this little guy came in very handy What are other data books that have impacted you and you believe are worth checking out?

I’ve just released a 6hr plus LangGraph fundamentals series all for free on YouTube (with practicals 😎) Find the link to the full playlist in the comments

@piggyvest what's with all the issues since last night. Can't do Safeflock, tried logging in on my desktop and it keeps navigating to this mobile page and asking me to log out (which also doesn't work by the way).

Another AI Agent Protocol is here 🚀 ......AG-UI [Full Video here youtu.be/Ho1tXkbV8Js] The future of AI isn't just about powerful models, but how intelligent agents interact and collaborate! Yet, building user-interactive AI agents and seamlessly connecting agent backends to frontend applications has been a significant challenge due to a lack of standardization. Introducing AG-UI: The Agent-User Interaction Protocol! This video dives into this new, open, and lightweight protocol designed specifically to standardize how AI agents connect to front-end applications. AG-UI acts as a streamlined bridge, enabling real-time communication and collaboration between your agent logic and the user interface. If you're building AI agents, multi-agent systems, or user-facing AI applications, understanding AG-UI is crucial. In this introduction, we will cover: - What AG-UI Protocol is: an open, lightweight, event-based protocol for agent-user interaction. - How AG-UI standardizes the connection between agent backends and front-end applications using a simple JSON event stream over HTTP. - Key concepts and features supported by AG-UI, including - The benefits of adopting a standardized AI agent protocol for development speed and interoperability. - How AG-UI complements other protocols like MCP and A2A. - AG-UI's compatibility with various agent frameworks (like @langchain , @crewAIInc ) and different backends. - Potential use cases for AG-UI powered applications. AG-UI establishes a consistent contract between agents and interfaces, simplifying development and unlocking new possibilities for human-agent collaboration. This video provides an essential introduction to this important new AI standard. 🔥 Be ahead of the curve! Learn about the new AI Agents Protocol - AG-UI and how it connects agents to users! 👍 If you see the value in standardizing AI agent-UI communication, hit the like button! 👇 What kind of interactive AI agents are you excited to build with a protocol like AG-UI? Share in the comments! 🚀 Stay updated on the latest in AI agents, protocols, and cutting-edge tech! Subscribe! 🔗 Share this video to spread the word about AG-UI and shape the future of AI agent interaction! Resources: - AG-UI Official Website & Documentation: docs.ag-ui.com/introduction or ag-ui.com - AG-UI GitHub Repository: github.com/ag-ui-protocol/ag… #AGUI #AIAgents #AIProtocol #ArtificialIntelligence #MultiAgentSystems #AIDevelopment #AIStandard #FutureOfAI #AgentCommunication #TechProtocol #Innovation #MachineLearning #DeepLearning #TechTutorial #Developer #AICoding #HumanAgentCollaboration #UserInterface #Frontend #Backend #Protocol

Full Lecture here youtu.be/QeEcUupMfFg

No Agents without Tool Calling [Link to Complete Video in the comments] Unlock the true power of Large Language Models! Are your LLM applications limited by static knowledge? Do you need your AI to interact with the real world, fetch real-time data, or perform actions? Base LLMs are powerful text generators, but their capabilities expand exponentially when they can use external tools. In this in-depth tutorial, we dive deep into Tool Calling and Function Calling within the @langchain framework. This is your key to building dynamic, capable AI applications that can browse the web, perform calculations, interact with APIs, and so much more! In this video, you will learn: - What "Tool Calling" and "Function Calling" mean in the context of LLMs and LangChain. - How to define and create simple tools in LangChain. - LangChain examples of using tools for real-world tasks. - Leveraging function calling APIs (like OpenAI's) within your LangChain applications. - Best practices for designing effective tools for your AI workflows. Understanding tool and function calling is crucial for any developer looking to build sophisticated applications with LangChain and Large Language Models. This video provides the practical knowledge you need to take your AI development skills to the next level and build intelligent agents that can perform complex tasks. 🔥 If you're building with LangChain, this video is a must-watch! 👍 Liked this tutorial? Give it a thumbs up! 👇 Have questions about Tool/Function Calling or LangChain? Leave a comment below! 🚀 Subscribe to the channel for more technical deep-dives into AI, LLMs, and development frameworks like LangChain!] 🔗 Share this video with anyone learning LangChain or AI development! #LangChain #ToolCalling #FunctionCalling #LLM #AIDevelopment #LangChainTutorial #LargeLanguageModels #AI #MachineLearning #Programming #Python #TechTutorial #AICoding #LangChainTools #IntelligentAgents #AIEngineering #DeveloperTutorial

A found a Gem, its called Gemini CANVAS It’s been out for a while but I never knew it was this awesome. With Gemini Canvas, you can directly interact with your results and fine-tune them in a dedicated editor instead of prompting for updated versions that fill up your chat session window and have you scrolling up and down to see what changed This is amazing especially for coding and writing articles You really need to check it out -------------------------------------------------- For more lessons, classes, tips and bits on AI Engineering, consider following me and subscribing to langcasts.com for high quality AI engineering content

Building an Agent that can Browse the Web The main power and USP of Agents is their ability to use tools to source external information and take external actions . I have been excited about the Google Agent Development Kit recently so I decided to build an agent that can search the web for real-time travel itinerary information like travel distance, weather information and many more. Find the full link to the tutorial in the comments -------------------------------------------------- For more lessons, classes, tips and bits on AI Engineering, consider following me and subscribing to langcasts.com for high quality AI engineering content

Let’s figure out Career Paths in AI (Full link in comments) The AI space is currently both exciting and overwhelming, however one thing is sure, there is great potential Now and Ahead. So how do you plug into this inevitable AI-powered future and finally cure your FOMO on AI Let’s figure it out in the next 20mins ----------------------------------------------------------- For more lessons, classes, tips and bits on AI Engineering, consider following me and subscribing to langcasts.com for high quality AI engineering content

Workflow Agents - Doing Agents the right way One concept I have fallen in love with in the new Google Agent Development kit is Workflow Agents. These types of agents allow you to gain full control over the activities of the autonomous agents in the system who can be highly non-deterministic. Workflow agents are not controlled by LLMs, instead, they are based on predefined logic. With workflow agents, you can predictably orchestrate the activities of the powerful autonomous ones, allowing them to run free under the supervision of their more controllable counterparts. I don’t know about you but I feel this, not extreme observability, will solve the problem of non-determinism in AI agents and enable engineers guide their agents to better produce expected results. You should check it out 😃 --------------------------------------------------------- For more lessons, classes, tips and bits on AI Engineering, consider following me and subscribing to langcasts.com for high quality AI engineering content

TALK to Your Agents and View Agentic workflows in the new Google ADK I started playing with the new Google Agent Development Kit some days ago and one of the things I find most exciting is the out-of-the-box agent testing UI. This UI allows you to interact with the agent you’re building immediately. It can even take multimodal inputs and one CRAZY part is that you can actually SPEAK to your agent. Oh yeah, you and your agent can have a one-to-one voice-based interaction Sure, agent visualization tools are not new, for example LangGraph studio. However, this is the easiest and most full-featured one I’ve seen yet You can also visualize your agent workflows, view events, access metadata, and track each activity taking place behind the scenes with your agents I’ll definitely be having a lot of fun with this in the coming weeks

Before you add A.I, fix this first... Dear product and SaaS companies, as you all raise to be a part of the AI era, it's important to continue to see features from the perspective of impact. Yes, we know you want to add that shiny new AI-powered feature, but why not prioritize fixing problems users have been complaining about for years first. Instead of Apple Intelligence, fix the magic mouse charging style Instead of launching 3000 Copilots, fix image alignment in Microsoft word While you invest billions of dollars into Anthropic, also invest a tiny fraction of that into improving the monumentally crappy AWS Console interface We don't always need AI, but we will always need solutions

Start with the Problem, not the AI I’ve been following Aishwarya Naresh Reganti's 10-day Agents email course (super amazing by the way) for some days now and this really stood out for me. While planning to integrate AI into your workflow, it’s common to make the mistake of trying to pick “the best Agent framework” or “your favourite Agent Architecture”. This foundational mistake can derail your agent adoption quickly. To avoid this early pitfall, consider the problem at hand, understand it’s requirements, then pick the Agent implementation type that best suites the problem you’re out to solve. Below is a cool guide that came with the email course to help you pick which agentic implementation strategy best suites your use case.

You’ll still need the same number of developers No. You won’t be replacing developers on your team, you will only be making the work of each one of them faster with A.I, and that is if you use A.I tools efficiently. This way, a deliverable that used to take 2 weeks, can now be delivered in a week. Sorry they promised you a lesser workforce, that everyone let go will be replaced by soulless robots and all you need to do is “hit play” while they get to work building your next billion dollar idea. That’s just all marketing speak. So, manage your expectations accordingly, because every work A.I augments; be it at the Frontend, Backend, DevOps etc., you’ll still need a developer steering it and also making sure it doesn’t actually get in the way of true productivity.

Start Learning About MCPs here Anyone with a skin in the AI Engineering game knows how much of a big deal MCPs are. I remember my first time learning Tool calling and I realised that as powerful as it was, we will need a harmonized standard for interacting with external tools, services and environments. And now we have MCPs, and while it still has its drawbacks at the moment, it’s gaining encouraging adoption (OpenAI recently) and that’s how you know that a technology has a great future. Now, I know you probably have been overwhelmed with so much “MCP posts”, some educative, others sensational. But if you want to learn without the noice, then I’ll recommended starting with these two materials: Anthropic’s announcement of MCPs: anthropic.com/news/model-con… The Official MCP Documentation: modelcontextprotocol.io/intr… (Bonus) A Curation of currently available MCP servers: github.com/modelcontextproto…

Before you build Agents, Learn these…. Agents are pretty cool ain’t they? That’s why we all want to build one But before you jump ahead with building an agent with LangGraph, CrewAI or the new OpenAI Agents SDK, go through the checklist below to see if you have already covered the core fundamentals of working with and building apps with LLMs to ensure that you don’t get frustrated in developing your agent - Prompt Engineering: For effective communication with your LLMs and within your agent hierarchies - Model Generation Configurations: The use of settings like temperature, max_output_tokens, rate limiting and topK, etc to gain control over how your LLM generates responses - Tool/Function Calling: Ability to extend LLM workflows with external functions and services - Embeddings: Encoding information semantically and in a format that LLMs can digest - Vector Databases: Storing information for semantic searches and long-term memory - Retrieval systems: Retrieving embedded information based on semantic similarity with outgoing prompts - RAG: Using external/private information to augment your LLMs knowledgebase - Multimodality: Working with different data formats asides text (audio, images, video, live voice etc) within your prompts and responses - Error handling: Mitigation strategies for handling errors within LLM applications - Orchestration: Ability to coordinate the interactions and flow of data for different components required for an LLM-powered application’s operations Once you have these basics covered, then you can confidently build agents without so much stress Subscribe to langcasts.com for more lesson on building AI Agents