Spacio’s Post

“𝗪𝗲 𝘀𝗵𝗼𝘂𝗹𝗱𝗻’𝘁 𝗴𝗶𝘃𝗲 𝘁𝗵𝗲 𝗰𝗿𝗲𝗮𝘁𝗶𝘃𝗶𝘁𝘆 𝗮𝘄𝗮𝘆. 𝗟𝗲𝘁 𝘁𝗵𝗲 𝗰𝗼𝗺𝗽𝘂𝘁𝗲𝗿 𝗱𝗼 𝘄𝗵𝗮𝘁 𝗶𝘁 𝗱𝗼𝗲𝘀 𝗯𝗲𝘀𝘁, 𝗮𝗻𝗱 𝘂𝘀 𝗵𝘂𝗺𝗮𝗻𝘀 𝗰𝗮𝗻 𝘁𝗮𝗸𝗲 𝗰𝗼𝗻𝘁𝗿𝗼𝗹 𝗼𝗳 𝘁𝗵𝗲 𝗰𝗿𝗲𝗮𝘁𝗶𝘃𝗲 𝗽𝗮𝗿𝘁.” We just published a recap of the amazing chat between Sofia from Parametric Solutions and Stian from Spacio at our Rebels of the #AEC event. They dove into the future of AEC software, and trust us, it’s a must-read! Here’s a quick peek: 𝗪𝗲𝗯 𝘃𝘀. 𝗗𝗲𝘀𝗸𝘁𝗼𝗽: Many design softwares shift from desktop to web-based tools, architects and engineers should expect quicker iterations from web-based solutions than what we have been used to getting from our desktop-based software. 𝗔𝗜 𝘃𝘀. 𝗛𝘂𝗺𝗮𝗻 𝗖𝗿𝗲𝗮𝘁𝗶𝘃𝗶𝘁𝘆: AI can handle the boring stuff, but the creative decisions? That’s still the responsibility of humans. 𝗖𝗼𝗹𝗹𝗮𝗯𝗼𝗿𝗮𝘁𝗶𝗼𝗻 & 𝗧𝗼𝗼𝗹𝘀: New tools are reshaping how we work together, and balancing control is a key consideration. Oh, and the audience asked a fun question: What animal is your software? Read the blog post here to learn the answer: https://lnkd.in/djA78xta #Innovation #AI #Architecture #technology

What if you could make the inside of the system present itself in a way that domain experts or business people could understand? Domain-Driven Design promoted the idea of a Ubiquitous Language as a means to bridge the communication gap between technical people and domain experts. People gather around a whiteboard and work on the design together. And then the names and logic are translated into the code. This works well, at least at the beginning. But how long can we maintain a manual view on the system in sync with the actual system. Not that long. But what if we can make the view be part of the system? What if, in addition to encoding the models and associated tests, we also encode the views from the whiteboard? Take a look at the attached little story for an example of how that can look like. It's about modeling prices with discounts by an amount of money. We start with a sketch on the whiteboard together with the domain experts. We end up with a picture showing a discount object that would behave like a price but internally wraps the original price. We like the idea and we implement the code including the associated logic to create a similar picture automatically. After a while we have a new idea. What if we represent the percentage discount in the same way. We get together with the domain experts and sketch on the whiteboard. But this time, we do not start from scratch. We start from the picture that the system already provides and we extend it manually with your extended idea. It turns out that generalizing to a composable decorator pattern might work nicely. We are happy about it and we implement it with proper tests (well, examples, but that's for another day), and, of course, without forgetting to endow our domain objects with the ability to draw themselves. And so on. For the entire system. And for each perspective we have about the system. At each point the increment is based both on the logic of the system and on the visualization of the present state. This visualization allows us to create a mental model of the system and enables us to reduce the communication imprecision. #MoldableDevelopment

𝗡𝗲𝘅𝘁-𝗚𝗲𝗻 𝗖𝗼𝗱𝗶𝗻𝗴 𝘄𝗶𝘁𝗵 𝗠𝘆 𝗗𝗮𝘂𝗴𝗵𝘁𝗲𝗿: 𝐒𝐨𝐟𝐭𝐰𝐚𝐫𝐞 𝐚𝐫𝐜𝐡𝐢𝐭𝐞𝐜𝐭𝐮𝐫𝐞! 𝐂𝐡𝐚𝐩𝐭𝐞𝐫 2: 𝐌𝐢𝐜𝐫𝐨𝐊𝐞𝐫𝐧𝐞𝐥 𝐀𝐩𝐩𝐫𝐨𝐚𝐜𝐡 Architecture design evolve beyond the classic multi-layered approach we described in chapter 1. While that model serves well in many scenarios, specialized needs call for different approaches. Exploring these alternatives reveals unique advantages, adapting to nuanced requirements. Our journey now shifts from the familiar to the innovative, uncovering architectures tailored for specific challenges. We built a similar application to chapter 1 image processing app but this time we applied various transformations to images: grayscale filtering, horizontal inversion and an automatic AI analysis (image to text). The application is designed using a 𝐌𝐢𝐜𝐫𝐨𝐤𝐞𝐫𝐧𝐞𝐥 𝐀𝐫𝐜𝐡𝐢𝐭𝐞𝐜𝐭𝐮𝐫𝐞, where the core system provides essential functionalities like the user interface, plugin management, and infrastructure services. Specific image processing features are implemented as separate plugins that can be dynamically loaded by the core application. This approach allows for easy extension of the application's capabilities, supporting new plugins that perform different image analyses or output types, such as text extraction. Overall, we can summarize the pros and cons like this: 𝐏𝐫𝐨𝐬: * Extensibility: New functionalities can be added as plugins without altering the core system. * Maintainability: The lightweight core reduces complexity, simplifying maintenance. * Isolation: Plugins function independently, enhancing system stability by preventing one failure from affecting others. 𝐂𝐨𝐧𝐬: * Complexity in Plugin Management: Managing dynamic loading and ensuring compatibility can be difficult. * Performance Overhead: Abstraction layers and inter-component communication may reduce performance. * Testing Challenges: Extensive testing is needed to ensure proper interaction between plugins and the core system. As usual, the code is open source and available on github (still work in progress, and here: https://lnkd.in/ecxp_iJX). Reminder: Chapter 1 about layered architecture is here: https://lnkd.in/et34QUwj

▪️Architect products. Evolve from projects to products.  Think of your work as a product that constantly changes, not as a static project. ▪️Focus on quality features, not just functional requirements. Quality features are the core of a sustainable architecture, like cost-effectiveness, configurability, performance, and robustness. ▪️Delay design decisions until absolutely necessary. Sometimes, design decisions are made when you don't know what the final product will look like. It's okay to start without a perfect solution that covers all possibilities, but be aware of creating "technical debt." ▪️Design for change—use the 'power of small.'  Changes are easier in smaller units than in large monoliths. However, don't make them too small to avoid excessive overhead. ▪️Design for building, testing, deploying and operating. Consider the impact on testing, deployment and operations. Not just the building process. ▪️Organize your teams to match the system design.  Set up teams in a way that fits the architecture of the systems they're working on. „Conway’s law“. (Written by human, formatted by AI) #enterprisearchitecture #architectureprinciples ---------- You like it? 🔔 Follow: Marcel Weigel ♻️ Share and let others benefit, too. 💾 Save for later.

In the Claude 3.5 work session that started here https://lnkd.in/eXVFV9hE , this document captures the development of the technical specification for the MGraph JSON Provider. **Part 1** - The Final Brief contains the current technical specification that we arrived at. This represents our simplified, graph-native approach that emerged through our discussions. **Part 2** - The Initial Design shows Claude's first solution. While comprehensive and solid, it followed more traditional patterns and had more complexity than needed. **Part 3** - The Design Journey documents our collaboration, including key questions, insights, and how we arrived at the simpler design. This project shows the value of thorough design work before implementation, but it also highlights an interesting shift that GenAI brings to technical documentation. In traditional development, documentation often becomes tech debt because updating it is seen as costly and time-consuming. Authors resist changes not based on whether they're right, but on how much work they'll create. With GenAI assistance, this friction disappears. Updating documentation becomes as natural as the design discussion itself. The focus stays where it should be: on finding the right solution, not managing documentation overhead. The result? A simpler, cleaner design that emerged through genuine architectural exploration, unhampered by documentation maintenance concerns. Update 1: Article about this workflow https://lnkd.in/eZvmE3_Y

Agent Design Patterns > Agent Framework Last week's post about framework documentation sparked an interesting debate. While having all those options is great, I've realized something more fundamental: What REALLY matters isn't the number of frameworks you know (actually the less the better) —it's understanding the core design patterns that make AI agents successful in production environments. Key considerations that actually move the needle: - Architectural trade-offs that align with business objectives - Production-ready patterns that scale - Design choices that support maintainable agent systems I've compiled an in-depth article on agent workflow patterns that focuses on some of these practical aspects. https://lnkd.in/gQm6pZPr Please note that this is a work in progress and I am more than happy to add your feedback #agent #genAI #design

Coming up! Part 2 of the 30-minute Domain Driven Design guide! Check out Part 1 here, and subscribe (for FREE) to get notified when part 2 is published (coming up next week). https://bit.ly/4d0syCW #DDD #Design #DomainDriven #SoftwareEngineering

📩 If you have applied to become a beta tester of our generative tool, check your email! In the email, you will find the information and user guide, along with early access to the software :) 🔐 We do not receive any data on your project, as we value data privacy. That's why your feedback will be valuable to us! 📞 If you have any questions, follow the contact form, write to us on the beta testers chat, or contact us directly on Linkedin. #aec #generative #architecture #software #building #city #architect #bim hashtag #3d #model

Did you miss this? The 30-minute guide to Domain Driven Design. Don't be limited by past technical decisions and preconceptions. Design the best possible future state. https://bit.ly/3zZ9sOG #ddd #design #technology

🏗️ 𝐄𝐥𝐞𝐯𝐚𝐭𝐞 𝐘𝐨𝐮𝐫 𝐀𝐄𝐂 𝐁𝐮𝐬𝐢𝐧𝐞𝐬𝐬 𝐰𝐢𝐭𝐡 𝐂𝐮𝐬𝐭𝐨𝐦 𝐒𝐨𝐟𝐭𝐰𝐚𝐫𝐞 𝐒𝐨𝐥𝐮𝐭𝐢𝐨𝐧𝐬! 💻 As a dedicated software developer in the Architecture, Engineering, and Construction (AEC) industry, I specialize in creating tailored applications that streamline tasks and optimize projects. 𝐒𝐚𝐯𝐞 𝐭𝐢𝐦𝐞, 𝐫𝐞𝐝𝐮𝐜𝐞 𝐜𝐨𝐬𝐭𝐬, 𝐦𝐢𝐧𝐢𝐦𝐢𝐳𝐞 𝐞𝐫𝐫𝐨𝐫𝐬, and 𝐛𝐨𝐨𝐬𝐭 𝐞𝐟𝐟𝐢𝐜𝐢𝐞𝐧𝐜𝐲 using my innovative solutions. 𝐌𝐲 𝐒𝐨𝐥𝐮𝐭𝐢𝐨𝐧𝐬: 🚀 𝐀𝐮𝐭𝐨𝐦𝐚𝐭𝐢𝐨𝐧: Streamline workflows by automating complex tasks, enhancing productivity, and allowing your team to focus on what matters most. 🔧 𝐄𝐫𝐫𝐨𝐫 𝐑𝐞𝐝𝐮𝐜𝐭𝐢𝐨𝐧: Minimize errors and enhance overall project performance with robust, reliable software solutions. 🤖 𝐀𝐈 & 𝐌𝐚𝐜𝐡𝐢𝐧𝐞 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠: Leverage cutting-edge AI and machine learning technologies for smarter decision-making and predictive analytics. 👁️ 𝐂𝐨𝐦𝐩𝐮𝐭𝐞𝐫 𝐕𝐢𝐬𝐢𝐨𝐧: Implement advanced computer vision techniques to gain deeper insights and improve accuracy in your projects. 🛠️ 𝐐𝐮𝐚𝐥𝐢𝐭𝐲 𝐂𝐨𝐝𝐞 𝐃𝐞𝐥𝐢𝐯𝐞𝐫𝐲: Ensure clean, maintainable source code with continuous integration and delivery (CI/CD) practices implemented on Microsoft Azure DevOps. ✅ 𝐋𝐞𝐭’𝐬 𝐂𝐨𝐧𝐧𝐞𝐜𝐭! If you're looking for custom software or consultation for your projects, feel free to reach out: 📧 xr.erz.apps@gmail.com 🔗 𝐄𝐱𝐩𝐥𝐨𝐫𝐞 𝐌𝐲 𝐖𝐨𝐫𝐤: Discover the plugins I’ve developed and published on the Autodesk App Store for Revit and Navisworks: https://lnkd.in/dxJNnUpv 🎞️ 𝐉𝐨𝐢𝐧 𝐌𝐞 𝐨𝐧 𝐘𝐨𝐮𝐓𝐮𝐛𝐞: Check out my channel for insights and tutorials: https://lnkd.in/dcTzsSdJ 🔖 𝐋𝐞𝐭’𝐬 𝐈𝐧𝐧𝐨𝐯𝐚𝐭𝐞 𝐓𝐨𝐠𝐞𝐭𝐡𝐞𝐫! Your next big leap in efficiency is just a conversation away. #SoftwareDevelopment #AEC #Automation #ProjectManagement #Efficiency #CustomSoftware #MachineLearning #AI #Architecture #Engineering #Design #Innovation #Autodesk #BIM #Plugin #Addin #Revit #Navisworks #AutodeskAppStore #Technology #CSharp #Construction #Azure #DigitalConstruction #Programming