Designing Interactive Prototypes

AI is rapidly moving from passive text generators to active decision-makers. To understand where things are headed, it’s important to trace the stages of this evolution. 1. 𝗟𝗟𝗠𝘀: 𝗧𝗵𝗲 𝗘𝗿𝗮 𝗼𝗳 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗙𝗹𝘂𝗲𝗻𝗰𝘆 Large Language Models (LLMs) like GPT-3 and GPT-4 excel at generating human-like text by predicting the next word in a sequence. They can produce coherent and contextually appropriate responses—but their capabilities end there. They don’t retain memory, they don’t take actions, and they don’t understand goals. They are reactive, not proactive. 2. 𝗥𝗔𝗚: 𝗧𝗵𝗲 𝗔𝗴𝗲 𝗼𝗳 𝗖𝗼𝗻𝘁𝗲𝘅𝘁-𝗔𝘄𝗮𝗿𝗲 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝗼𝗻 Retrieval-Augmented Generation (RAG) brought a major upgrade by integrating LLMs with external knowledge sources like vector databases or document stores. Now the model could retrieve relevant context and generate more accurate and personalized responses based on that information. This stage introduced the idea of 𝗱𝘆𝗻𝗮𝗺𝗶𝗰 𝗸𝗻𝗼𝘄𝗹𝗲𝗱𝗴𝗲 𝗮𝗰𝗰𝗲𝘀𝘀, but still required orchestration. The system didn’t plan or act—it responded with more relevance. 3. 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝗔𝗜: 𝗧𝗼𝘄𝗮𝗿𝗱 𝗔𝘂𝘁𝗼𝗻𝗼𝗺𝗼𝘂𝘀 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 Agentic AI is a fundamentally different paradigm. Here, systems are built to perceive, reason, and act toward goals—often without constant human prompting. An Agentic system includes: • 𝗠𝗲𝗺𝗼𝗿𝘆: to retain and recall information over time. • 𝗣𝗹𝗮𝗻𝗻𝗶𝗻𝗴: to decide what actions to take and in what order. • 𝗧𝗼𝗼𝗹 𝗨𝘀𝗲: to interact with APIs, databases, code, or software systems. • 𝗔𝘂𝘁𝗼𝗻𝗼𝗺𝘆: to loop through perception, decision, and action—iteratively improving performance.    Instead of a single model generating content, we now orchestrate 𝗺𝘂𝗹𝘁𝗶𝗽𝗹𝗲 𝗮𝗴𝗲𝗻𝘁𝘀, each responsible for specific tasks, coordinated by a central controller or planner. This is the architecture behind emerging use cases like autonomous coding assistants, intelligent workflow bots, and AI co-pilots that can operate entire systems. 𝗧𝗵𝗲 𝗦𝗵𝗶𝗳𝘁 𝗶𝗻 𝗧𝗵𝗶𝗻𝗸𝗶𝗻𝗴 We’re no longer designing prompts. We’re designing 𝗺𝗼𝗱𝘂𝗹𝗮𝗿, 𝗴𝗼𝗮𝗹-𝗱𝗿𝗶𝘃𝗲𝗻 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 capable of interacting with the real world. This evolution—LLM → RAG → Agentic AI—marks the transition from 𝗹𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝘂𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱𝗶𝗻𝗴 to 𝗴𝗼𝗮𝗹-𝗱𝗿𝗶𝘃𝗲𝗻 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲.

🍱 How To Design Effective Dashboard UX (+ Figma Kits). With practical techniques to drive accurate decisions with the right data. 🤔 Business decisions need reliable insights to support them. ✅ Good dashboards deliver relevant and unbiased insights. ✅ They require clean, well-organized, well-formatted data. ✅ Often packed in a tight grid, with little whitespace (if any). 🚫 Scrolling is inefficient in dashboards: makes comparing hard. ✅ Start with the audience and decisions they need to make. ✅ Study where, when and how the dashboard will be used. ✅ Study what metrics/data would support user’s decisions. ✅ Explore how to aggregate, organize and filter this data. ✅ More data → more filters/views, less data → single values. 🚫 Simpler ≠ better: match user expertise when choosing charts. ✅ Prioritize metrics: key insights → top left, rest → bottom right. ✅ Then set layout density: open, table, grouped or schematic. ✅ Add customizable presets, layouts, views + guides, videos. ✅ Next, sketch dashboards on paper, get feedback, iterate. When designing dashboards, the most damaging thing we can do is to oversimplify a complex domain, or mislead the audience. Our data must be complete and unbiased, our insights accurate and up-to-date, and our UI must match users’ varying levels of data literacy. Dashboard value is measured by useful actions it prompts. So invest most of the design time scrutinizing metrics needed to drive relevant insights. Bring data owners and developers early in the process. You will need their support to find sources, but also clean, verify, aggregate, organize and filter data. Good questions to ask: 🧭 What decisions do you want to be more informed on? (Purpose) 😤 What’s the hardest thing about these decisions? (Frustrations) 📊 Describe how you are making these decisions? (Sources) 🗃️ What data helps you make these decisions? (Metrics) 🧠 How much detail is needed for each metric? (Data literacy) 🚀 How often will you be using this dashboard? (Value) 🎲 What constraints should we know about? (Risks) And, most importantly, test dashboards repeatedly with actual users. Choose key tasks and see how successful users are. It won’t be right at first, but once you get beyond 80% success rate, your users might never leave your dashboard again. ✤ Dashboard Patterns + Figma Kits: Data Dashboards UX: https://lnkd.in/eticxU-N 👍 dYdX: https://lnkd.in/eUBScaHp 👍 Ethr: https://lnkd.in/eSTzcN7V Orange: https://lnkd.in/ewBJZcgC 👍 Semrush: https://lnkd.in/dUgWtwnu 👍 UKO: https://lnkd.in/eNFv2p_a 👍 Wireframing Kit: https://lnkd.in/esqRdDyi 👍 [continues in comments ↓]

📱 Designing for Thumbs 🧠 Here’s how users hold their phones: • 49% use one hand • 36% cradle the phone (supporting with one hand, interacting with the other) • 15% use two hands This makes a huge difference in how we design mobile interfaces. If key actions aren’t within easy thumb reach, we create unnecessary friction. Here’s a quick breakdown of the thumb zone: 🟢 Comfort zone – effortless, natural access 🟡 Stretch zone – reachable, but requires effort 🔴 Risk zone – awkward or frustrating to tap If your primary actions sit in the red — your design is likely causing frustration. ✨ Great mobile UX isn’t just clean — it’s comfortable. 💡 I always position primary actions in the green zone — especially navigation, CTAs, and core gestures. Small shifts here make a big difference in usability. Let’s keep bridging the gap between beautiful and usable. 📚 Recommended reads: The Thumb Zone – Scott Hurff - https://lnkd.in/dQyzEjBq One-Handed Mobile Use – Luke Wroblewski - https://lnkd.in/dg5g3QMZ #UXDesign #MobileDesign #UIDesign #ThumbZone #ProductDesign #MobileUX #Figma #DesignThinking #Accessibility #HumanCenteredDesign

Claude Design can do some very interesting things for FP&A teams. I tested it with a practical use case: communicating a forecast to leadership. Watch it here: https://lnkd.in/esNfsaVX From a messy FP&A output and turning it into something decision-ready. If you want dataset and prompts so you can follow along, tell me in the commments! Step 1: Clean the data in Claude I uploaded a messy forecast dataset and prompted Claude to structure it. It normalized the format, fixed inconsistencies, and generated a clean forecast table along with a data quality log. Step 2: Validate and refine I reviewed the audit log and refined the dataset. The goal here is not perfection, but having consistent, structured data that can be used for reporting. Step 3: Build the FP&A dashboard in Claude Design I uploaded the cleaned dataset into Claude Design and prompted it to create an executive dashboard. I defined KPIs, time horizon, and variance views. The result was a multi-tab dashboard with visuals and built-in commentary, already usable for leadership. Step 4: Create the CFO presentation Using the same analysis, I prompted Claude Design to generate a board-style presentation. It structured the story into an executive summary, financial performance, variance drivers, and recommended actions without rebuilding everything manually. Step 5: Turn it into a video briefing Finally, I asked it to convert the presentation into a short video-style briefing. It produced a concise narrative that explains what changed, why it changed, and what leadership should focus on next. What stood out is that Claude Design is not just a design tool. It acts as a translation layer between financial analysis and executive communication. Instead of rebuilding the same story across Excel, slides, and summaries, you can move from analysis to dashboard to presentation to narrative in one flow. This is where a lot of FP&A time is still being lost today. Not in the analysis itself, but in turning that analysis into something leadership can actually use.

🖐️✨ Who needs a mouse when you can just wave your hands dramatically in the air? Just wrapped up my latest project: AirCanvas - a computer vision application that lets you draw using just hand gestures! AirCanvas transforms your webcam into an interactive digital canvas using: • MediaPipe for real-time hand tracking • OpenCV for image processing • Computer vision techniques to interpret gestures as drawing commands 🎮  How it works: • 🤏 Pinch your index finger and thumb to draw • ☝️ Point with your index finger to select colors • 🖐️ Show an open palm to erase 🧠 What I learned: • MediaPipe is surprisingly accurate for hand tracking • The challenges of creating intuitive gesture recognition systems • That waving your hands around in public coffee shops leads to some... interesting conversations with strangers ("No, I'm not summoning spirits, just trying to draw a circle") This project was a fun exploration of creating natural human-computer interfaces. 👉 Check out the code and try it yourself: https://lnkd.in/ebmF28ky #ComputerVision #OpenCV #MediaPipe #GestureRecognition #Python

One of the easiest ways to improve your collaboration with product managers and engineers is to improve your design handoffs. - It doesn’t take a lot of time - It drastically improves how people interpret your designs - It reduces backs and forths between engineers and designers - It gives engineers confidence that they are doing the right thing - It increases the chances that what’s implemented will match the design Great handoffs, for me, are an instant sign of design maturity. It shows me that designers think not only of themselves but also the ecosystem of people around them. Coming up with great handoffs boils down to the following: - Setting context - Adding structure - Adding annotations - Including all states - Visualizing the flow - Including a prototype - Doing a run-through In the age of AI, including a high-fidelity AI prototype (with Figma Make, Replit, Lovable or any tool of your choice), can also drastically improve how well engineers understand your intentions, particularly with complex flows or interactions. Not everything is required for every design initiative; a small feature update may not need a full handoff, whereas a big, impactful one may require everything from the above. In this cheat sheet, I’ve put together my top tips for delivering the “perfect” handoff. Bonus: I have also created a Figma Annotation and handoff kit with handy components for the above. Find the link in the comments. 👇 — If you found this useful, consider reposting ♻️ #uxdesign #uiux #productdesign

What's the worst investor advice for an early-stage founder? “Build an MVP.” Most startups follow it, and then regret it. If you are pre-launch, what you really need is a Minimum Viable Test (MVT). A MVT is a focused experiment designed to learn about the riskiest assumption. If you're worried about technical feasibility, write the smallest script or prototype that tests just that. If your biggest uncertainty is user demand, try a landing page test with five users. If logistics is your weak point, run a single transaction manually and see what breaks. You don’t need a product. You need evidence. Confidence to build comes from reducing uncertainty, not shipping a half-done app. That’s how I work with early-stage teams now. We run small tests, fast. We ask, “What’s the least we can do to learn the most?” Sometimes that means a simple Figma mockup, rather than a working app. Sometimes it’s a spreadsheet, rather than an analytics dashboard. Sometimes it’s just some terminal output, rather than an interactive UI. We run test after test, killing ideas quickly or leveling them up with real insight. Once the biggest risks are solved, building the full product becomes obvious. Don’t start with an MVP. Start with a test.

Obsessing over the perfect prompt only takes you so far. The key to building AI agents that actually work in production? 𝗖𝗼𝗻𝘁𝗲𝘅𝘁 𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴. Why it matters: - LLMs don’t have infinite attention. - Every extra token eats into a finite “attention budget.” - At some point, too much context causes “context rot”, where the model forgets or confuses the very thing you wanted it to recall. Anthropic 𝘀𝗵𝗮𝗿𝗲𝗱 𝘀𝗼𝗺𝗲 𝗲𝘅𝗰𝗲𝗹𝗹𝗲𝗻𝘁 𝗯𝗲𝘀𝘁 𝗽𝗿𝗮𝗰𝘁𝗶𝗰𝗲𝘀 𝗳𝗼𝗿 𝗰𝗼𝗻𝘁𝗲𝘅𝘁 𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴: 1/ Start with a minimal but clear, structured system prompt. 2/ Provide few, well-designed tools that are token-efficient & unambiguous. 3/ Use a few canonical examples, not exhaustive lists of edge cases. 4/ Use “just-in-time” context (loading only what’s needed dynamically) instead of frontloading all data. 5/ Summarize and refresh context or persist memory outside the model's window for long-horizon tasks and continuity. As models improve, they need less handholding. But context engineering will remain essential. It ensures agents stay coherent, efficient, and effective, especially in long, complex tasks. For more info, check out the original report from Anthropic in comments ↓ #EnterpriseAI #AIAgents #AIforBusiness

When I was 14, I sold a product that wasn't real. On purpose. I wanted to start a mail-order business selling fly-tying materials to fishermen. But I had no idea if anyone would actually buy. So I placed an £8 advert in Trout & Salmon magazine: "Send for my catalogue." The problem was, I hadn't printed the catalogue yet. I hadn't even bought any stock. When 25 people responded, I told them we had "sold out" and they were out of print. Then I scrambled to put one together. That £8 test told me everything I needed to know. There was demand and the business was viable. I went on to turn over £1,500 in the first year, with £356 profit. That felt good for a teenager with a £100 loan from his mum. Here's what I learned about validation: ➡️ Test before you invest The biggest mistake founders make is building before they validate.  They spend months (sometimes years) perfecting a product nobody wants. ➡️ Make your test affordable £8 bought me the answer to a £10,000 question.  You don't need venture capital to test an idea.  You need creativity and nerve. ➡️ Make your test fast I had my answer in a week. That's how I discovered that speed matters.  The longer you wait to test, the more attached you become to an idea that might not work. ➡️ Let the market decide I didn't ask friends what they thought.  I didn't run focus groups.  I put real money on the line and saw the results. ➡️ Copy what works, then improve it I didn't invent fly-tying materials.  I just found a better way to sell them.  Take what's already working and find a way to execute it better. It's about getting it 80% right, then letting your customers show you the rest. The software industry worked this out years ago.  They release version 1.0 knowing it's not perfect. Then they improve based on real feedback. You can do the same, whatever your business is. A simple test you can run this week: Before you invest a large amount of money, run the smallest possible test that proves demand. - A classified advert like I did. - 10 conversations with potential customers. - A prototype made from cardboard and duct tape. Whatever proves people will actually pay for what you're planning to build. Because the market will always tell you the truth if you're willing to ask. If you're currently testing a business idea,  I'd like to hear how you're validating demand before you build. 

A college student with quadriplegia types math equations with her tongue. No surgery. No headgear. Just a retainer. She’s studying math and computer science. She takes notes in class, texts friends, and researches assignments—often in places where voice control isn’t an option. Libraries. Lectures. Group study rooms. The device is MouthPad^, built by MIT spin-off Augmental. It’s a custom 3D-printed retainer that sits on the roof of your mouth and turns tongue movement into cursor control. Think about what assistive tech usually asks people to accept: ↳ Surgical implants, recovery time, and real risk ↳ Visible headgear that turns you into “the person with the device” ↳ Voice commands that fail the moment the room needs to be quiet ↳ Price tags that put independence out of reach What Augmental built instead: ↳ A touchpad for the tongue: swipe to move, press to click ↳ Head-tracking for broad motion, tongue reserved for precise control ↳ Bluetooth pairing like a normal mouse—no special software ↳ $1,500 during early access ↳ Users wearing it up to 9 hours a day with low fatigue Here’s the part that stopped me. Her mother called receiving the MouthPad^ “the most significant moment since her injury.” Not the diagnosis. Not the rehab milestones. A dental retainer that gave her daughter control back—without making her life louder, more visible, or more medical. In a lecture hall, no one knows she’s controlling her laptop. That’s not a side benefit. That’s the design. For boards and executive teams, this isn’t a feel-good edge case. It’s governance. When non-invasive, low-cost accessibility exists and organisations still choose systems that shut people out, the issue stops being “technology limits” and becomes leadership. Procurement standards, inclusion policy, and digital decisions determine who gets full access—and who is left working around choices made for them. That’s when accessibility stops being a design preference and becomes an accountability issue. And it raises a blunt question: Why did we spend decades trying to restore digital access by opening skulls when a workable interface might already be inside the mouth? ♻️ Share if you think accessibility shouldn’t require surgery—just better design. This is one of the stories from my keynote on Designing Accessible Futures: AI, Ethics and Inclusion in Practice for boards and leadership teams. Source: MIT News and Augmental (MouthPad^ assistive technology). Video Insta luminica.ai