Cubode’s Post

Do you want to build an AI Agent, but don't know where to start? We are building an AI Agent, and we are documenting it all. With full articles over on Medium that explain every step of the process and how to build it yourself! We have covered things like setting up the correct AI environment, picking our front-end "framework", boosting inference speed so now we are covering our back-end framework. We have just finished a 5 part article series of How to set up Django from Scratch with Celery + Channels + Redis + Docker. And we talk through everything you need to know including the structure to build the agent in the back end, and how to set up each framework with DJango! I will leave the links in the comments to each of the articles, go build your own!

Integrating LANGCHAIN with GROQ! Ever seen Al Magic? 🤖 This is when it happens 🪄 Creating a super-efficient Al Agent requires power for fast, accurate responses.💬🏎️ We want our Agent to autonomously generate manually customisable data visualisation plots from an uploaded CSV file. So naturally speed and reliability were our non-negotiables. We integrated LangChain with Groq, tested with Llama3 70B & Mixtral 8x7B. Results? Responses in under 2 seconds! 😱 Want the nitty-gritty details? Comment "Link" for the FULL coding notebook and results to see for yourself to learn how you can optimise your Al agent's performance with LangChain and Groq! #A| #Tech #LangChain #Groq #FastAl #llama3 #mixtral

This will make your AI product the fastest in the market. When building our AI Agent, speed and reliability were non-negotiables, so we have chosen Groq! What is Groq? It's an AI infrastructure designed for fast AI inference. We're utilizing their Language Processing Unit (LPU), a new type of processing system that offers the fastest inference for computationally intensive applications like Large Language Models (LLMs). I will drop the full article in the comments for the full coding notebook, if you want to integrate Groq yourself or just want to see how we did so go have a read!

More than 200 people have read this article in the last 24h 🤯 Are you going to miss it? We showed how we used Pydantic data models to turn unstructured responses from an LLM into outputs that we can reuse more efficiently and integrate with other parts of our system. This is part of our 30 day challenge to build an AI Agent that auto-generates manually customisable plots in public. Link in comments 🔗 👇

How is ChatGPT so fast? Faster than any human ever?🤔 It’s all about inference speed! 🚀 Inference speed is the backbone of Al applications. Its what makes chatbots quick, translations seamless, and Al interactive. For LLMs like GPT4, faster inference means real-time responses. 🤖💬 I will leave the full article with all the extra info in the comments, go take a read!

What is AI most in need of? Determinism. The What You See Is What You Get kind of determinism, that's what this mini AI Agent we're building brings to plotting data. AI will do the heavy lifting and you will be able to change plot type, colours, titles, legend, aggregation type... In a fully interactive plot that's auto generated when you load your CSV directly on your browser. This is the latest update of an ongoing product, not available yet, but you can join the waiting list if you want to give it a go, link in comments. #GenAI #AIAgents #AgenticWorkflows #DataVisualisation #Echarts

How to Build an AI Agent: Picking our Plotting Library If you don't know how to build an AI Agent, like most of us! We are building one so you can learn with us! The task for the day was to choose a library for the plots the Agent would generate. We chose ECharts, an incredible open-source JavaScript library that allows for fully cutomisable and interactive charts that live in your browser! We did look at alternatives like Vega-Altair and Plotly, but ultimately we decided on ECharts for its versatility and simplicity. Let me know if you’ve used this library yourself!

Is Django the Best Backend Framework for AI Products? Hi, this is Ben 👋 The team asked me to explain why we used Django for the backend of our AI Agent challenge, in under 30 seconds! I utterly think it's the best choice, I've left the full article with a comparison analysis against Flask and FastAPI in the comments. It's one of our most read articles 😎 We work in bets, with strong beliefs weakly held, so if you have proof that other frameworks are best, I'd love to hear from you in the comments or DMs #ai #aiagent #codechallenge #backend #django #aichallenge

We choose Langchain to build our first AI Agent and wrote an in depth analysis of why we think Langchain is better than LlamaIndex for this project, with 5 different tests with and the steo-by-step code tutorial (link in comments), including RAG and agent build itself. Langchain gives you simplicity and speed of development that LlamaIndex can't offer. This is day 5 of our 30 day challenge to build an AI Agent, open sourced and documented. #langchain #ai #aiagents #aiagent