Articles of the Week
From Flutter in crisis to developers outsourcing their brains to AI
Hey everyone! How's it going? This week, our dev brains were buzzing with a load of super interesting articles we found. We decided to share some summaries of what we've been reading with you - it ranges from concerns about the future of Flutter to the controversies about the new generation of programmers who only know how to use AI. Check it out and let us know what you think!
Flutter is at a critical inflection point
The author, a long-time Flutter fan, expresses concerns about the future of the framework due to Google's lack of focus, questionable product choices and poor communication. He argues that the decision to use C++ for the desktop platform implementation was a mistake, and that the lack of communication about changes to the formatter caused frustration in the community.
Lew C laments Flutter's wasted potential on the desktop and fears that a lack of priority and poor communication could alienate developers and harm the future of the framework. It's a shame to see such a promising framework facing these challenges, but the author remains hopeful that Flutter can overcome these obstacles and continue to be a great option for application development.
Building a RAG-Powered Support Chatbot in 24 Hours of Hackathon
Akshay Gupta tells us about his team's adventure at a hackathon, where the challenge was ambitious: to create an AI support chatbot that thinks and responds like an experienced platform administrator. Using large language models (LLMs) and caffeine, they created a surprisingly capable RAG system.
The team deserves all the praise for taking on this code marathon! Creating a functional chatbot in just 24 hours is impressive, especially using cutting-edge technologies like LLMs and RAG. If the result was good enough to win the hackathon, imagine what they could do with more time and a few good coffees!
The ChatGPT Hack Top 1% Developers Use to Write Code 10x Faster
If you're one of those who ask ChatGPT to "write a function that...", you can stop now! The people who are dominating the use of AI for programming aren't just copying and pasting from a souped-up Stack Overflow. The secret is to use ChatGPT as a senior architect, discussing the system design in depth, implementing it little by little, and asking for code review. That way, instead of just being a "code monkey", you turbocharge your brain with decades of experience.
I found the idea of using ChatGPT as an architecture partner, and not just as a code generator, interesting. It seems that the trick is to know how to give the right commands to AI and use its knowledge to think big, instead of just solving small problems. And of course, without forgetting that we have to review everything, right? After all, not even AI is perfect!
Comparing local-first frameworks and approaches
This article explores local-first frameworks and approaches, highlighting how they are reshaping data ownership and application architecture. It covers everything from the concept of CRDTs for efficiently merging data to the use of PouchDB for data synchronisation and ElectricSQL for Postgres synchronisation.
Local-first isn't just about cutting-edge technology; it's about reimagining existing databases for a world where offline is the norm. Putting data back in the hands of users is a clever move for more resilient, privacy-respecting and user-centred applications.
The MongoDB Mistake That's Costing Your Startup Millions
This article by Mehdi Bafdil addresses common MongoDB configuration errors that can lead to high costs and performance issues. The author explains how N+1 queries, unoptimised queries, a lack of concern for data integrity and the absence of indexing can negatively impact performance and increase spending. In addition, he offers practical tips and a checklist to optimise MongoDB and avoid these problems.
If you're using MongoDB, this article is a find! It gives you a heads-up about the most common mistakes that could be costing you a fortune and tanking your application's performance. With the tips and checklist, you'll turbocharge your MongoDB and save some money!
I Dropped SQL for NoSQL. Our App Now Handles 5x the Traffic
The article reports the experience of a company that switched its SQL database for a NoSQL one, MongoDB, to handle the increase in application traffic. The decision was made after frustrated attempts to optimise the SQL database, which was causing slowness and errors. The migration to MongoDB resulted in significant improvements in application performance, as well as reduced infrastructure costs and boosted team morale.
I thought the company's attitude in switching from SQL to NoSQL was very brave, but it seems like it paid off, right? After all, who doesn't want a faster, cheaper application with happy people? 😉 We love Mongo DB here at Yes Marketing.
Beyond Visual: Why We Should Be Using More Haptic Feedback on the Web
Lucas de França reminds us that the web is very visual, but it could be more tactile. He explains how haptic feedback can improve the user experience, making everything faster and more intuitive. And to make life easier for devs, he even created a React hook called useVibration.
I thought the point raised was super valid. We forget that touch can be a great tool on the web. And Lucas's hook seems to be a godsend for people who want to experiment with it. Will we soon have sites that "vibrate" when we click on something? 🤔 Here at Yes we've even brainstormed about the topic. And another thing, Lucas created a hook for React, but you can use the Vibration API with any web library or framework!
Agentic AI and the MCP Ecosystem
The article discusses the evolution of AI to agent systems, which not only respond to prompts, but also take actions and interact with external systems. It explains how these agents work with LLMs and the importance of "tool calling" to access and execute APIs. The article also introduces the Model Context Protocol (MCP) as a solution for integrating agents with various systems dynamically, and mentions Goose as an example of an extensible AI agent.
It's amazing how AI is evolving! Before we were just chatting, now AI is solving our programming problems. The article is very complete and explains exactly how everything works, from "tool calling" to the so-called MCP. It's worth taking a look if you want to understand the future of AI and apply it to your projects.
Young Coders Are Using AI for Everything, Giving "Blank Stares" When Asked How Programs Actually Work
The youngsters of the new generation of programmers are using artificial intelligence for everything, and when we ask how things really work, they draw a blank! Like, they ask for help from AI even to breathe, and then we wonder: do these people really know how to program?
If this continues, soon we'll have to ask AI to fix its own errors, because no one will know how things really work anymore. But hey, at least we'll have "fast" code, right?
They lied to you. Building software is really hard.
The article addresses a reflection on the promise of low-code tools and artificial intelligence (AI) in accelerating software development, arguing that, while they may be useful for rapid prototyping, they can limit long-term growth and the ability to solve complex problems.
We know very well that the journey to becoming a real software developer is full of challenges and constant learning. Running away from this path may seem tempting at first, but in the end, it will only take you further away from your goal. So, are you going to face it or are you going to be fooled by false promises?
And that's it, folks! We hope you enjoyed this selection of readings as much as we did. We realised that there's a lot of discussion going on about AI in programming - from how to use it best to the dangers of relying on it too much. We also have the classic database debates and the challenges of popular frameworks. Until the next post with more hot tips from the dev world! 👨💻✌️
