Author: Alfred Nutile

Date: 06/05/2025

Watch the Video

Okay, so this video is all about Convex Chef, an open-source AI coding agent that helps you spin up full-stack, real-time apps using just a single prompt. No backend setup, no limits, apparently! It leverages the Convex reactive database and throws in some cool features like “Recipes” for instantly adding things like real-time text editors or even AI chat.

Why is this inspiring? Well, for those of us knee-deep in transitioning to AI-enhanced development, this is a potential game-changer. Think about it – we’re moving away from hand-cranking every line of code and towards orchestrating AI to build the scaffolding for us. The video shows Convex Chef building clones of complex apps like Notion and Slack, all from a single prompt, which is mind-blowing. Plus, with Gemini 2.5 Pro integration for AI-powered coding and a built-in real-time backend, it sounds like a full package.

Imagine using this to prototype a new feature for a client in a matter of hours instead of days. You could iterate faster, experiment with different UIs, and focus on the core business logic instead of wrestling with infrastructure. The open-source aspect is a huge plus, too. It means we can peek under the hood, contribute, and potentially customize it to fit our specific needs. Honestly, the prospect of building a functional app with essentially a text prompt is too tempting to ignore. I’m downloading it now to give it a spin – who knows, it could become a staple in my AI-powered toolkit.

Date: 06/04/2025

Watch the Video

Okay, this Supabase MCP Server video is seriously cool, and here’s why I think it’s worth your time. It shows how to give your AI agent deep context about your Supabase project, essentially letting it “understand” your backend in the same way it groks file structures and code. Jon Meyers walks through setting up the Supabase MCP server within Cursor IDE and then uses Claude to whip up an Edge Function that intelligently scrapes recipe websites. Forget the ad-ridden, SEO-spam versions – this pulls out just the core recipe data and stores it in a Postgres database, then displays it in a Next.js app.

The real value for us, as developers moving towards AI-assisted workflows, is how it streamlines development and automation. Imagine the possibilities! Instead of manually writing complex scrapers and data cleaning scripts, you can leverage AI to handle that heavy lifting. I’ve spent countless hours wrestling with web scraping in the past (and honestly, who hasn’t?), so seeing this level of automation makes me genuinely excited. This isn’t just about scraping recipes; it’s about connecting AI to your database schema, table relationships, and even your custom functions, allowing it to assist in tasks you hadn’t even imagined.

I’m already brainstorming ways to apply this to our internal tools and client projects. Think automated data migrations, intelligent report generation, or even AI-powered API development. This video gives a practical, hands-on example of how to bridge the gap between LLMs and real-world development tasks. The combination of Supabase’s backend capabilities and AI coding tools like Claude could seriously boost productivity and unlock new levels of automation, it’s definitely worth experimenting with.

Date: 06/04/2025

Watch the Video

Okay, so this video is all about leveraging Flowise v3 to build AI agents using a no-code interface. It shows you how to create an assistant that can do everything from searching the web and answering questions from uploaded documents to connecting with external APIs like Gmail. It also dives into setting up document stores, handling data chunking with OpenAI embeddings, and integrating Postgres vector databases (Supabase). The real kicker is the focus on practical tools like SERP API and Composio to truly extend what your AI agent can do.

Why is this video gold for us? Because it’s a tangible step towards blending traditional development with AI-powered automation. We’re constantly looking for ways to reduce boilerplate and speed up development cycles. Seeing how Flowise v3 simplifies the creation of sophisticated AI agents with zero code is incredibly appealing. Imagine being able to prototype and deploy a complex workflow in a fraction of the time, without getting bogged down in the nitty-gritty of code.

Thinking about real-world applications, this opens doors to things like automated customer support systems that can pull information from various sources, or even intelligent data processing pipelines that can analyze documents and trigger actions in other applications. The Gmail integration alone has me thinking about automating email workflows based on document content. I’m personally excited to experiment with Flowise to create a personalized research assistant that automatically aggregates information from different sources and summarizes it for me. The video’s clear, step-by-step approach makes it feel immediately accessible, and honestly, that’s half the battle when diving into new tech!

Date: 06/03/2025

Watch the Video

Okay, this Flowise v3 video looks seriously useful, especially given the direction I’m pushing my own workflow! Essentially, it’s a deep dive into building AI-powered apps using a no-code platform. Think visual flowcharts connecting LLMs (like OpenAI), custom knowledge bases, and even human-in-the-loop steps. It’s positioned as a real alternative to tools like N8N and Zapier, putting AI capabilities front and center.

The reason this is valuable for someone like me (and potentially you) is that it bridges the gap between traditional coding and leveraging the power of AI. We can use our existing knowledge of APIs and system design, but visually orchestrate complex AI interactions without writing tons of boilerplate code. Imagine building an automated customer support system powered by a custom-trained knowledge base without spending weeks on intricate Python scripts. This video walks you through the basics from local installation to creating your first AI agent, handling memory, state, debugging and more!

What’s got me excited is the potential for rapid prototyping and experimentation. Instead of getting bogged down in the weeds of low-level implementations, I can focus on the logic of the AI workflow. Building AI agents and integrating them with other tools becomes something that is actually doable. I’m definitely going to spin up a local instance and see how quickly I can adapt some of my existing projects. The promised easy integration with custom APIs and the document store functionality are huge wins too! Honestly, for a developer looking to embrace the AI wave, this seems like a solid place to start playing.

Date: 06/03/2025

Watch the Video

Okay, this AG-UI video looks seriously interesting, especially for how I’m aiming to streamline AI integrations into my projects. In essence, it’s about AG-UI, an open-source protocol to create frontends for AI Agents that you can stream events to—messages, typing indicators, the whole shebang. Think of it as a bridge connecting your LLM-powered agent’s brain to a slick, real-time UI.

Why does this pique my interest? Well, I’ve been wrestling with clunky, custom solutions for displaying AI interactions. This video walks through setting up a FastAPI backend that’s AG-UI-compatible and then connecting it to a CopilotKit frontend. So imagine moving beyond basic API calls and actually streaming the AI’s “thoughts” as they happen—no more stale, static responses! This aligns perfectly with the AI-enhanced workflows that I’m migrating to.

The real kicker is the potential for automating UI development around AI agents. Instead of hand-crafting chat interfaces every single time, AG-UI provides a standardized protocol. I’m already envisioning how I can adapt this to some of my existing Laravel projects that use LLMs for content generation and customer support. Instead of wrestling with SSE implementation myself, I can leverage a ready-made solution. I’m gonna dive into this!

Date: 06/02/2025

Watch the Video

Okay, this video on Supabase Edge Functions on Fly.io is gold for any of us transitioning to AI-driven workflows. Jon Meyers walks through deploying a Supabase Edge Function, essentially a serverless function, on Fly.io using Deno and Oak middleware. This means we can ditch some of the heavier backend lifting and focus on orchestrating logic with tools like LLMs.

Why’s it valuable? Because it showcases how to self-host these functions, giving us control and flexibility. Instead of being tied to a specific cloud provider’s serverless platform, we can deploy these lightweight functions anywhere, including environments where we’re integrating AI agents or no-code solutions. Imagine using an LLM to generate the core logic within the Edge Function, then deploying it to a cost-effective and scalable platform like Fly.io, orchestrated entirely by AI. We can have AI write the function, write the tests, and orchestrate the deployment!

The real-world application is huge. Think automated content generation, dynamic API endpoints, or even real-time data transformation triggered by AI models. By experimenting with this, we’re not just learning about Supabase or Fly.io; we’re building a foundation for a whole new level of automation and intelligent applications. It’s definitely worth carving out an hour to play with!

Date: 06/02/2025

Watch the Video

Okay, this video on self-hosting Supabase with S3 storage, custom domains, and Authelia is exactly the kind of thing I’m diving into! It’s a walkthrough of setting up a complete backend infrastructure, and what’s killer is the focus on self-hosting. We’re talking full control, reducing reliance on external services, and potentially big cost savings down the road. It’s not just about slapping together a quick prototype; it’s about building a robust, production-ready environment.

What makes this video inspiring is that it bridges the gap between traditional backend setups and the newer, “serverless” world that Supabase offers. The inclusion of Authelia for authentication shows a real-world security mindset. We, as devs, can leverage the techniques shown here to move away from the complexity of frameworks such as Laravel and use Supabase as BaaS and build an entire scalable app using Vue, React or Svelte (my favourite). The video even acknowledges some initial hiccups (which the author immediately fixed), it adds a layer of authenticity.

I’m already thinking about how I can use this setup for a client project where data sovereignty and control are paramount. Instead of relying on a managed Supabase instance, I can deploy this on a Hetzner or DigitalOcean server, giving the client complete ownership of their data. This video is a must-watch for any developer looking to level up their backend game and explore the power of self-hosted solutions, and the mentioned hiccups only add credibility. I’m going to experiment with this over the weekend.

Date: 06/02/2025

Watch the Video

Okay, so this video is all about tackling a major pain point in RAG (Retrieval Augmented Generation): keeping the knowledge base up-to-date. It introduces Graphiti, an open-source platform for building real-time knowledge graphs, and demonstrates how it supercharges your RAG agents. Think of it as moving from static textbooks to a living, breathing, constantly updated encyclopedia for your AI.

For me, as someone knee-deep in transitioning to AI-powered workflows, this is gold. We’ve all been there, right? Spending hours syncing data sources with our RAG systems, knowing it’s inefficient and prone to errors. The promise of a dynamic knowledge graph that evolves with your data is HUGE. The video not only explains the concept but provides a quickstart guide and a demo, making it really practical. The comparison with other RAG strategies like GraphRAG/LightRAG, along with Agentic RAG, is icing on the cake.

Imagine using this to build a dynamic product catalog for an e-commerce site, or a real-time support system that learns from every interaction. The possibilities are endless. I’m particularly excited about the agentic RAG part, where the AI agent can actively update the knowledge graph based on its interactions. This video is definitely worth experimenting with because it addresses a core challenge in RAG and offers a practical, open-source solution that could save tons of time and unlock new levels of AI-driven automation. I’m already thinking of ways to integrate this with my Laravel projects!

Date: 05/31/2025

Watch the Video

Okay, this week’s AI roundup video is seriously speaking my language! It’s all about leveraging free and open-source tools to tackle tasks that used to either be expensive or rely on proprietary platforms. We’re talking about AgenticSeek for local web automation (a potential Manus AI replacement!), Resemble AI’s Chatterbox for on-premise, high-quality text-to-speech, and Flux Kontext Pro dramatically cutting image generation costs.

What makes this video so valuable for us developers diving into AI is that it showcases practical alternatives to the big, well-known AI services. Forget being locked into expensive APIs! AgenticSeek, for instance, offers a chance to build custom web scraping or task automation workflows without relying on third-party services, giving us more control and potentially better privacy. Chatterbox hits home because having a solid, locally hosted TTS model opens doors for accessibility features, voice-based UIs, or even dynamic content generation directly within our apps. And that 86% cost reduction for image generation? That’s HUGE! Think about automating marketing materials, generating custom thumbnails (like they do with “Pimp My Thumb” in the video!), or even prototyping UI elements without breaking the bank.

For me, the Flux Kontext Pro segment is particularly exciting. Imagine integrating it into a Laravel app where users can generate custom images based on data they input. Suddenly, we’re not just building web apps; we’re building AI-powered creation platforms. It’s absolutely worth experimenting with these tools, even if there’s a learning curve. The potential for automation, cost savings, and unlocking new features in our projects is just too compelling to ignore.