Collaborative Design Environments

MCP vs A2A vs ACP vs ANP Let’s start with the 'solutions', when to use which: ✅ Use 𝗠𝗖𝗣 when you want your model to be context-aware ✅ Use 𝗔𝟮𝗔 𝗽𝗿𝗼𝘁𝗼𝗰𝗼𝗹 when your system needs multiple agents to collaborate, each with their own role and responsibility ✅ Use 𝐀𝐂𝐏 when your agents need to communicate, coordinate, or negotiate with structured intent ✅ Use 𝐀𝐍𝐏 when you want a scalable way for agents to discover and connect across a distributed system Now, time for their stories!✨ ➝ 𝗠𝗖𝗣 (𝗠𝗼𝗱𝗲𝗹-𝗖𝗼𝗻𝘁𝗲𝘅𝘁 𝗣𝗿𝗼𝘁𝗼𝗰𝗼𝗹) In short, it’s a way to give LLMs a structured understanding of the world around them: who the user is, what tools are available, what memory to retain. It’s context-as-code, not just context-as-prompt. (I shared more in my last post, I’ll drop the link in the comments) ➝ 𝗔𝗴𝗲𝗻𝘁-𝘁𝗼-𝗔𝗴𝗲𝗻𝘁 𝗣𝗿𝗼𝘁𝗼𝗰𝗼𝗹𝘀 This is about how multiple agents communicate with each other, not just the user anymore. Instead of one big model doing everything, you break the task into smaller parts handled by different agents. They take on roles, pass tasks, and coordinate to solve more complex goals. In most current systems, this is done through direct message passing, often one agent at a time, with fairly simple, turn-based logic. The structure is usually custom and manually defined, it's effective, but still early-stage. That’s why more standardization is starting to emerge, to make these systems more modular and scalable. ➝ 𝐀𝐂𝐏 (𝐀𝐠𝐞𝐧𝐭 𝐂𝐨𝐦𝐦𝐮𝐧𝐢𝐜𝐚𝐭𝐢𝐨𝐧 𝐏𝐫𝐨𝐭𝐨𝐜𝐨𝐥) This defines how agents talk to each other. Not just sending messages, but structuring intent: → Is it a request, a proposal, or an update? → What shared terms or logic do they rely on? ACP is like a shared language for agents, so they can collaborate, negotiate, and reason together. ➝ 𝗔𝗡𝗣 (𝐀𝐠𝐞𝐧𝐭-𝐍𝐞𝐭𝐰𝐨𝐫𝐤 𝐏𝐫𝐨𝐭𝐨𝐜𝐨𝐥) Now structure comes in. Not just in what agents say, but in how they connect. While A2A describes the idea of agents collaborating, ANP defines the transport: how agents discover each other, route messages, and coordinate across systems. It’s the backbone that makes reliable, scalable agent communication possible. ****They’re all 𝗯𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗯𝗹𝗼𝗰𝗸𝘀: - MCP provides structured context - A2A enables role-based collaboration - ACP defines how agents communicate - ANP brings order to large-scale agent networks The layered design is what powers the most capable AI systems today: context-aware, action-taking, and agent-driven by design. 📍If you are into building/learning Agents, don't miss this AI Agent hackathon for noncoders (with $5000 prize!): https://lnkd.in/dkdxyhD9 You’ll choose a real public-sector challenge and use AI to create a practical solution. You won’t be building a toy app. You’ll build something you can demo. And defend. Before the hackathon begins, we run prep sessions inside GenAI.academy so you’re not starting from zero. Have fun!🐣

Walk anywhere in Amsterdam and you’ll sense a calm coherence: bricks, curbs, fixtures, benches, bollards and drains that look related. This isn’t accidental. It’s a product of the Puccini Method, the city’s standards—part design language, part technical playbook—for shaping every street and square. Adopted as citywide policy in 2018, it defines how the public realm is designed: from choice of pavers to lighting, furniture, tree species; even details like gullies and edging. The aim is streets that are functional, durable, safe, and visually consistent, without tipping into fussy “over-design”. Puccini emerged to fix two chronic problems: visual clutter and procurement patchwork. Before the framework, boroughs sourced their own elements, leaving a jumble of styles and standards. A single method delivers economies of scale, easier maintenance and—crucially—calmer, more legible urban spaces. In practice, Puccini produces vanzelfsprekend (or "self-evident") streets. The palette favours restrained forms and finishes, with familiar Amsterdam cues. This quiet consistency reduces visual noise, helps people navigate, and simplifies upkeep for crews who know exactly which component goes where. The method also hardwires sustainability into everyday decisions; tying procurement and design to environmental criteria, encouraging durable materials, repairable components and circular approaches. In short: long-life surfaces, robust furniture, and planting that can thrive as the climate changes. That sustainable spine is reinforced by “Green Puccini”: citywide agreements for the quality and management of planting. The dedicated handbook details tree species, ground covers, soils, and maintenance, aligning biodiversity and climate resilience with the same rigour given to bricks and lighting. The Puccini Method doesn’t chase spectacle. Its power lies in thousands of ordinary, repeated decisions that add up to a city that feels legible, durable and unique. In an era of eye-catching urban design, Amsterdam’s approach is refreshingly modest: design the everyday well, and do it the same way everywhere that makes sense.

🎩 “How We Designed a Multi-Brand Design System” (https://lnkd.in/erc3mA4i), a fantastic case study by Ness Grixti on the pains of maintaining multiple systems, how to make a design system work seamlessly across multiple brands — with a multi-system token infrastructure in Figma, applied everywhere. Most teams eventually go through a “consolidation” effort — and that’s where struggles emerge as different systems have slightly different needs. Wise team created a system where the entire library is controlled by a top brand layer, which contained nested libraries for type, spacing and colors. I love Wise's approach of 1 system with 2 tracks to avoid duplications — typography and spacing on one track, and color on the other. Both pulled from shared primitives and brought together in a single Global Token library that houses all brands. As a result, designers can manage responsive type, spacing and interaction states across all color themes from one place. 🧱 Raw values core → for color, type, spacing without brand associations 🧅 Layered structure → primitives, scaling/device, sentiment, brands 🌈 Sentiment themes → for alerts, neutral, warning, success, proposition 📏 Accessible dynamic scaling → for type and spacing values 📦 Nested variables → scaling lives within responsive device library ♻️ Avoiding token explosion → tokens shared across brands + diffs If you'd like to dive deeper, I highly recommend to take a look at the Multi-Brand Design System Figma Kit (https://lnkd.in/eShgnPnW) by Pavel Kiselev, a practical guide and Figma kit on how to set up a design system in Figma for multiple brands, platforms or products — with full control over colors, typography and visual styles. And huge kudos to Ness Grixti (along with colleagues Henrique Gusso and Willem Purdy and the wonderful Wise team) for sharing the challenges, the failures and successes for all of us to learn from! 👏🏼👏🏽👏🏾 #ux #design

Imagine a product you couldn’t use alone. Sounds impractical. Until you see what it unlocks. Coca-Cola once redesigned its bottle so the cap needed two people to open it. Not faster. Not easier. Together. Launched in 2014 on Colombian college campuses, each bottle featured a specially engineered cap with half of a interlocking twist mechanism…think puzzle pieces that only fit together. A single person couldn't open it alone; two bottles had to align precisely and twist simultaneously to unlock both caps, forcing collaboration in seconds. They had to look up, find someone nearby, align themselves, and twist in sync. That small interruption changed everything. The bottle stopped being just a beverage and became an excuse. To smile. To speak. To cross that invisible line between strangers. What appeared playful on the surface was deeply intentional underneath. Without a single line of copy, it engineered connection. This is design at its most powerful. Not shouting for attention. Not piling on innovation for the sake of it. This addressed freshman isolation amid rising phone dependency, transforming a routine sip into an interactive challenge. It embodied Coke's "share happiness" ethos not through ads, but by embedding connection into the product itself. Rooted in social design, it prioritized emotional connection over utility, addressing college students' screen-induced loneliness. By design, it created micro-moments of vulnerability…asking for help build rapport instantly, fostering trust and conversation. Videos of these moments went viral, amassing 9 million YouTube views and doubling sales. This human-centric approach boosted brand loyalty by solving relational needs. It raises an important question for anyone building products, brands, or systems today: What would happen if we designed less for convenience and more for shared human moments?

Design as Water, treating water as stakeholder in urban projects Reposting a handbook that still stands out: Design as Water by Henning Larsen and Ramboll. It is a practical guide that helps urban teams shift from “manage water” to “design with water”. The core idea is simple but demanding: water is not only a resource or a risk to control. It is a stakeholder with its own logic, needs, and time horizons. The beautifully designed guide combines images and text seamlessly, creating an engaging resource for practitioners seeking transformative approaches. The Urgency is Clear The climate crisis has intensified water volatility dramatically. Since the 2010s, flooding and heavy rainfall frequency has increased by over 50%, with water-related disasters comprising 90% of all natural disasters. By 2050, half the world's population is expected to live in water-stressed areas. Current rigid water systems are poorly equipped to adapt to these changes. The conventional paradigm of controlling water - forcing it within artificial boundaries and treating it as an exploited resource - has reached its limits. Water as Stakeholder This companion, developed through a co-creative process with numerous contributors, challenges professionals to reimagine water not as a resource to manage, but as a vital stakeholder with its own voice and needs. The guide offers practical changes for urban projects: Project Evaluation: Incorporating water impact as screening criteria, asking "What if water was our client?" when challenging project briefs. Team Integration: Appointing water stewards and integrating water experts who understand local water dynamics. Site Understanding: Looking beyond project boundaries to grasp interconnected water systems - watersheds, groundwater, and connections to larger water bodies. Long-term Vision: Designing for water's needs over 10-500 years, anticipating climate impacts whilst restoring historical interventions that constrained flows. Practical Implementation Rather than controlling water, the approach advocates giving water space through natural flows, green infrastructure, and permeable surfaces. The guide embodies water's qualities - connected, responsive, transparent, ever-evolving, playful - suggesting meandering processes rather than linear approaches. For planners, developers, and policymakers, this represents designing resilient urban futures. Water doesn't need humans, but humans cannot survive without water. #watedesign #climateadaptation #urbandesign #spatialplanning #bluegreeninfrastructure #resilience #cities

𝗔𝗜 𝗮𝗴𝗲𝗻𝘁𝘀 𝗮𝗿𝗲𝗻’𝘁 𝗺𝗮𝗴𝗶𝗰! 𝗧𝗵𝗲𝘆’𝗿𝗲 𝘀𝗼𝗳𝘁𝘄𝗮𝗿𝗲 — 𝗮𝗻𝗱 𝘁𝗵𝗲𝘆 𝗻𝗲𝗲𝗱 𝗮𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 𝗷𝘂𝘀𝘁 𝗹𝗶𝗸𝗲 𝗲𝘃𝗲𝗿𝘆𝘁𝗵𝗶𝗻𝗴 𝗲𝗹𝘀𝗲 𝘆𝗼𝘂 𝘀𝗵𝗶𝗽. ⬇️ If you’re building for retrieval-intensive use cases — copilots, automated analysts, async workflows — you need more than a clever prompt. You basically need a solid agentic architecture. 𝗟𝗲𝘁'𝘀 𝗯𝗿𝗲𝗮𝗸 𝗶𝘁 𝗱𝗼𝘄𝗻: ⬇️ An AI agent isn’t just a chatbot. It’s a runtime system that coordinates and contains the following components: ➜ Reasoning – Breaking down tasks, choosing the right tools, adjusting the plan ➜ Memory – Using past interactions, history, and long-term context ➜ Tools – Dynamically calling APIs, databases, search engines, or other agents This is the real shift: From prompt engineering to systems engineering. But it doesn't stop at individual components. The real power of agents emerges when these elements are combined — and when multiple agents operate together with structure and intent. 𝗦𝗼 𝘄𝗵𝗮𝘁 𝗶𝘀 𝗶𝘁 𝗮𝗯𝗼𝘂𝘁 𝗦𝗶𝗻𝗴𝗹𝗲-𝗔𝗴𝗲𝗻𝘁 𝘃𝘀 𝗠𝘂𝗹𝘁𝗶-𝗔𝗴𝗲𝗻𝘁: ⬇️ Single-agent architectures are simple. Great for linear tasks. Easy to debug ( you just take the above components). But they fall short when things get complex — multi-step workflows, conflicting constraints, dynamic inputs. That’s where multi-agent architectures take over. Multiple agents, each with a specialized role, working in coordination. Think of it as microservices — but for cognition. 𝗖𝗼𝗺𝗺𝗼𝗻 𝗣𝗮𝘁𝘁𝗲𝗿𝗻𝘀 𝗶𝗻 𝗠𝘂𝗹𝘁𝗶-𝗔𝗴𝗲𝗻𝘁 𝗦𝘆𝘀𝘁𝗲𝗺𝘀: ➜ Parallel: Multiple agents working at once  Use case: Splitting document parsing or code review ➜ Sequential: Output from one becomes input for the next  Use case: Multi-step approvals, or structured reasoning ➜ Loop: Agents revisit tasks iteratively until confidence is high  Use case: Testing, code validation, recursive thinking ➜ Router: A central decision-maker assigns tasks to the right agent  Use case: Workflow orchestration based on context ➜ Network: Agents collaborate in a peer-to-peer model  Use case: Decentralized planning, brainstorming ➜ Hierarchical: A manager-agent supervises sub-agents  Use case: Enterprise-grade decision trees, command chains If you're aiming for scale, reliability, and true autonomy, architecture isn't optional. The future of intelligent systems will not be defined by a single, monolithic agent attempting to do everything. It will be shaped by coordinated systems of specialized agents, each designed for specific roles, operating together like a well-structured organization. Kudos to Weaviate for this great visualization!

Living near your best friends after college isn't immature. It's the smartest housing decision you'll make. You need to build for it, not age out of it. Once you do, you unlock community most people only dream about. Here's how Phil Levin turned this idea into $70M in pipeline: Most people think living near friends is something you age out of after senior year of college. Phil Levin thought that was backwards. So he gathered 20 adults and kids to live within walking distance. Not a commune or co-living. Just intentional proximity. Now he's scaling it with Live Near Friends. Phil's been writing about this idea since 2020. For three years, there was modest traction. Then the audience exploded. His theory: the pandemic made everyone rethink how they wanted to live. Remote work isolates us. Screen time distracts us. And yet, we’re craving the opposite. But we’re all too busy to figure out the “how.” The validation came fast: A single TikTok post drew 13,000 comments. Then the money: $70M in projects in the Bay Area in a matter of weeks. Instead of writing a business plan, Phil lived it. He spent four years testing the model with his family. Proof of concept, then he built the company. Two groups want this badly: • Young families (dual-income, time-strapped, dreading the suburbs alone) • Women 55-75 with equity who want real community and connection Both are hungry for housing that starts with people, not the property. Live Near Friends isn't about giving up privacy. You keep your own home. You're just choosing to be near the people you'd call at a moment's notice. That changes everything: • Kids play outside freely • Parents share the load • Dinners happen spontaneously Real belonging. The village model, reimagined. Phil's learned something important. Real estate can build real brands. And those brands translate directly into returns. But the brand has to stand for something real. Staged photos don't work. The founder has to embody it. Phil literally lives with his friends. That personal story resonates far more than any marketing. Phil spoke at Blueprint on the innovation stage. We discussed how creative real estate models solve human problems, not just building problems. The most disruptive play isn't always technology. Sometimes it's just remembering that people need people.

Agentic systems are transforming the way we build intelligent applications. But building one that scales 𝘳𝘦𝘭𝘪𝘢𝘣𝘭𝘺 requires more than just chaining prompts or APIs. It demands a robust architecture — one that blends structure, adaptability, and memory. Here’s a sketch I created to summarize a complete 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝗔𝗜 𝗕𝗹𝘂𝗲𝗽𝗿𝗶𝗻𝘁, inspired by real-world systems: Core Components 1. 𝗟𝗟𝗠 (𝗟𝗮𝗿𝗴𝗲 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗠𝗼𝗱𝗲𝗹) – The foundation for reasoning, communication, and synthesis. 2. 𝗣𝗹𝗮𝗻𝗻𝗶𝗻𝗴 𝗔𝗴𝗲𝗻𝘁 – Creates task decomposition and selects optimal workflows.     3. 𝗘𝘅𝗲𝗰𝘂𝘁𝗶𝗼𝗻 𝗔𝗴𝗲𝗻𝘁𝘀 – Operate in:    →𝗦𝗲𝗾𝘂𝗲𝗻𝘁𝗶𝗮𝗹 𝗪𝗼𝗿𝗸𝗳𝗹𝗼𝘄 (agent → agent handoff)    →𝗣𝗮𝗿𝗮𝗹𝗹𝗲𝗹 𝗪𝗼𝗿𝗸𝗳𝗹𝗼𝘄 (simultaneous agent execution with a Decision Agent) 4. 𝗚𝘂𝗮𝗿𝗱𝗿𝗮𝗶𝗹𝘀 – Ensure ethical, safe, and bounded operations (PII protection, response filtering, etc.) 5. 𝗠𝗲𝗺𝗼𝗿𝘆 𝗠𝗼𝗱𝘂𝗹𝗲𝘀 – Capture and use:    Chat History    User Profile    Conversation State 6. 𝗢𝗯𝘀𝗲𝗿𝘃𝗮𝗯𝗶𝗹𝗶𝘁𝘆 & 𝗔𝗻𝗮𝗹𝘆𝘁𝗶𝗰𝘀 – Track performance, bottlenecks, and system drift.     Frameworks That Map to This Architecture This blueprint isn't theoretical — it's actionable with the right tools: • 𝗟𝗮𝗻𝗴𝗚𝗿𝗮𝗽𝗵 → Graph-based stateful agent flows • 𝗖𝗿𝗲𝘄𝗔𝗜 → Autonomous teams of specialized agents • 𝗔𝘂𝘁𝗼𝗴𝗲𝗻 (𝗠𝗶𝗰𝗿𝗼𝘀𝗼𝗳𝘁) → Conversational agent orchestration framework • 𝗠𝗲𝘁𝗮𝗚𝗣𝗧 → Multi-agent system for software generation • 𝗔𝗗𝗞 (𝗔𝗴𝗲𝗻𝘁 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁 𝗞𝗶𝘁) → Brings modularity, plug-and-play memory, observability, and execution logic to life.    Each of these fits naturally into this architecture — some emphasize planning, others coordination or tooling — but 𝘁𝗵𝗲𝘆 𝗮𝗹𝗹 𝘀𝗵𝗮𝗿𝗲 𝗮 𝗰𝗼𝗺𝗺𝗼𝗻 𝗴𝗼𝗮𝗹: 𝗯𝘂𝗶𝗹𝗱 𝘁𝗿𝘂𝗹𝘆 𝗮𝘂𝘁𝗼𝗻𝗼𝗺𝗼𝘂𝘀, 𝗮𝗱𝗮𝗽𝘁𝗶𝘃𝗲 𝘀𝘆𝘀𝘁𝗲𝗺𝘀.

Multi-agent systems don’t fail loudly. They fail pathologically. Three failure modes I keep seeing in production: 1️⃣ Deadlocks Agent A waits for Agent B. Agent B waits for Agent A. Neither times out. System looks “healthy.” Nothing moves. 2️⃣ Thrashing Agents repeatedly undo each other’s actions. Planner rewrites. Executor reverts. Loop continues. Tokens burn. No progress. 3️⃣ Feedback Amplification One agent generates uncertainty. Another agent “clarifies.” First agent reinterprets clarification. Confidence decreases with every iteration. These aren’t model problems. They’re coordination failures. If you’re building multi-agent systems, you need: 1. Explicit state ownership 2. Progress guarantees 3. Hard iteration caps 4. Conflict resolution rules Orchestration is a systems problem. Not a prompting problem. What’s the worst emergent behavior you’ve seen in agent systems? #AIEngineering #MultiAgentSystems #AgentArchitecture #DistributedSystems #SystemDesign #AIOrchestration #ReliabilityEngineering #ProductionAI #EmergentBehavior #EnterpriseAI

"We stopped talking about return to office and started talking about reattaching." — Ryan Anderson, MillerKnoll Stress and burnout continue to grow and building engagement at work has taken a distant back seat to the continued drive for efficiency. Recent Upwork research reveals a troubling trend around AI: heavy users are becoming emotionally disconnected from their teams -- they actually trust AI more than their colleagues. What if, instead, we took some of that time back and invested in relationships? As Ryan put it "looking at AI as a way of reinvesting time savings in more relational human activities." The solution isn't just getting bodies in seats. It's designing spaces that strengthen human relationships. His team at MillerKnoll has identified what works in "relationship-based design": 🏢 Cafes with intention: Different table heights and seating arrangements that give people "permission to go meet someone new"—from quick corridor intercepts to intimate booth conversations. 📺 Meeting spaces for equity: Moving away from "Death Star-like" conference rooms to inclusive spaces where everyone has clear sight lines, whether remote or in-person. 🚪 Private offices reimagined: Designs that invite people in rather than create power distance—even executive offices can build relationships if you're intentional. Anderson's insight: successful workplace design is "50% space, 50% engagement." If people understand that a space is designed to help them connect and learn from each other, they'll actually use it that way. 👉 Read on for more in-depth #workplace design research: https://lnkd.in/d6fDvugg How are you designing your workplace to strengthen relationships, not just support tasks?