Maggie Appleton

The design team at Octopus Deploy has been looking at ways to modernize the UI. We’ve released an update for the process editor, a critical part of where teams build and maintain their deployments The redesign reduces clutter, visually groups parent and child steps (which customers have repeatedly asked for), and gives the UI a modern refresh. It’s intended to reduce friction, help customers move faster, and give them greater control over complex processes. This is the first of quality of life UI improvements to come 😄 For now, read more at https://lnkd.in/gf6zKNnz

40

4 Comments

Unsurprisingly the feeds have been ablaze with talk of Figma's MCP server, and yesterday Akbar Mirza joined Liam Hampton to discuss some good practices to ensure a more accurate output in the code generation inside of VSCode. One of the key takeaways was the importance of documentation in your co-pilot instructions.md file, suggesting fixes based on some of the recurrent issues. It gives me huge confidence that the labour intensive process of Design System and Developer documentation is truly starting paying off. You can check out the full session here ⤵️ https://lnkd.in/gavm-T_8

40

Developing a Shared Language for Designers and Engineers by Mandy Michael Lost in translation between Figma and production? Mandy Michael addresses the age-old problem of designer-developer miscommunication with modern solutions that actually work. At Octopus Deploy, they've moved beyond traditional handoffs to create a genuinely shared vocabulary that both designers and engineers understand. This session demonstrates how tools like Figma Code Connect ensure design tokens translate perfectly to CSS variables, how MCP servers let AI understand both design intent and code constraints, and how documentation becomes a bridge rather than a barrier. Through practical examples, learn to establish naming conventions both disciplines embrace, create component libraries that serve both design and development, and build workflows where changes in design automatically update development references. Learn more and register now for Dev Summit 2025, taking place in Sydney and online November 2025!

1

2 Comments

We hosted the Nile x Anthropic x WTW Neuron panel last week (read as: Sam Irving, Adriaan and I sat on a couch for an hour and chatted about AI) We focused on AI in regulated businesses - what's actually working, what's harder than expected, where it's heading. Against the backdrop of Gemini 3 and Opus 4.5 being released, it was a good time to compare notes! Some bits I'm still thinking about: The junior role question is one we need to take seriously. Traditionally, entry-level work has often been grunt work - the stuff that teaches you the ropes through repetition. If AI takes that away, what does a first role actually look like? How do we develop people deliberately rather than through osmosis? The tech is moving faster than most organisations can absorb. The bottleneck is rarely the AI itself - it's culture, structure, permission to experiment. If the technology stopped improving tomorrow, we'd still see massive gains year on year just from companies learning how to use what already exists. And if you're designing around today's AI limitations - hallucination, context, memory - without thinking about how fast those are being solved, you're probably aiming too low. Thanks to Lucy Barrett for keeping us on track, and to Hampden Bank for hosting us! Some of the sharpest points came from audience engagement on the day - so opening it up here too. Anything to add or challenge?

51

Designer to Problem Solver with Creative Freedom? For real? I'm very excited to have Pir Ahmed join me on The Prodcast! He’s currently a Product Designer at PureSquare and the Founder of Desnify, where he’s helping purposefully build design systems and scalable product experiences. In this episode, we’ll talk about: 🛠️ Starting Desnify and leading design independently 🔁 Scaling repeatable systems for startups 💡 What young designers should stop doing when applying for jobs Pir’s hybrid experience as both a startup founder and a product designer makes this episode a goldmine of insights! 🎧 Stay tuned for the full conversation. To know more about me: YouTube: @asadnawazzzz Instagram: @asadnawazzzz Website: asadnawazzzz.com #ProductDesign #Podcast #Designdiscussion #Designerthoughts

17

4 Comments

I’ve been thinking a lot about how the designer role is changing in practice, especially as AI becomes part of day-to-day work. Less time on one-off screens and more time working closer to code. Defining tokens, variables, systems, and components, while AI helps with the heavy lifting around scale and consistency. #ProductDesign #DesignSystems #UXDesign #DesignEngineering #AIinDesign #ProductTeams https://lnkd.in/eb-6mFDW

9

The best infrastructure is boring and lasts forever. 
When the foundational HCI and core interaction models are reliable and invisible, applications layered on top can finally breathe — giving room for real creativity.

4

At Big Human, we've been seeing what we can build with Lovable. I wrote about some of those experiments here: https://lnkd.in/eWMxSF_r TLDR: I think there's finally a real opportunity for people to start making their own bespoke tools and the barrier to entry is extremely low.

6

1 Comment

Design systems have always had an adoption problem. You build components. Document them. Publish the library. Then someone uses the right button in the wrong context. Skips the confirmation on a destructive action. Builds a flow that technically uses the system but it doesn't really work with the rest of your app. Sound familiar? Same components, different judgment calls. Now multiply that by AI. Designers in Cursor. Engineers in Claude Code. Each prompt starts from zero. Same component library but totally different decisions about behavior. I've been experimenting with specs that load automatically before anyone prompts anything. Not docs people have to remember to read. Context that's just there in the tools you're using. Tested this solo, most recently with Try/Keep, which went from design to App Store. Now exploring how it scales with teams at Sibi. That's why I'm publishing: to learn from the community what breaks at scale. The article covers what goes in the specs and starter prompts to try it yourself. If you're doing something like this with your team, I'd love to hear what's working and what's not. https://lnkd.in/gEnKBgGb #buildinpublic #designsystems #aitools #aifordesigners #productdesign

12

Entitlements are the building blocks of monetisation infrastructure. Getting them right matters. At MonetizationOS we've been thinking hard about the mental models behind entitlements and how to make them work for everyone, not just engineers. I wrote about how we solved it. Link in the comments. ⚙️

20

3 Comments

AI tools for design systems have matured quickly. The operational work hasn't kept up. The token audits. The deprecation plans. The governance documentation nobody reads until something breaks. The stakeholder brief you put together at 11pm before a quarterly review. The drift you fight quietly before it becomes someone else's emergency. I've been using Claude Code skills to handle that work for a while now – structured prompts that know how to read a token file, classify drift, produce a deprecation plan with blast radius analysis, or translate system health data into business language. I tidied them up and put them somewhere public in case anyone else finds them useful. The first public release is live, and I'm expecting gaps to surface as more people run them against real systems. I'll be pushing updates as they do. https://lnkd.in/dBxPazYu

123

10 Comments

Google just launched Gemini 3, its most capable model yet - and I’ve been genuinely impressed. I waited to try it myself before forming an opinion, and what struck me most wasn’t the reasoning or speed - it was how confidently it could design. Real, usable interfaces. Good hierarchy. Even decent typography. And that made me pause. For years I’ve said that design isn’t about moving rectangles in Figma. It’s about thinking: flow, clarity, intent, emotion. Yet, for every new designer entering the field, those rectangles still matter - they’re how you learn to see. Now I’m wondering: what happens when you can skip that part? Maybe in a year or two, junior designers won’t start by pushing pixels. They’ll start by prompting, by shaping flows in words, not layers. Just like architects once drew plans by hand before software took over, the next generation might jump straight into directing outcomes instead of tracing lines. And that’s both thrilling and strange. Thrilling - because it elevates the role of the designer from maker to conductor. Strange - because we may lose a piece of the craft that taught us to feel proportions, to see balance, to understand detail. AI isn’t erasing design. It’s redefining the entry point. Maybe the rectangles will stay - maybe they’ll just live one layer deeper, invisible under the prompts. Either way, design will still be about the same thing: understanding humans and shaping systems that fit them. The tools just keep evolving.

20

1 Comment

🚀 This week I tested how far Figma Make can go while building Roberts Intelligence - a lightweight project & time-tracking micro-SaaS. 🧠 Key learning: Figma Make is fantastic for rapid UI prototyping, but it cannot create or use real Supabase tables. It works only with a predefined key-value store, regardless of the relational schema you request. A great tool for ideas, flows, and quick UX, but not enough for production-ready data or real backend validation. 📖 Full story here: https://lnkd.in/dXz9x-um #ai #figmamake #supabase #microSaaS #productdesign #uidesign #uxdesign #saasdevelopment #startupbuilders #productstrategy #prototyping #designtech #noCode #builders #robertsintelligence

1

There are a lot of provocative takes right now about the future of design. “Figma is dead.” “It’s all Claude.” “It’s all Stitch.” Most of these are thinking in extremes. The reality is simpler: Pick the right tool for the job. In exploration, you optimize for speed. In validation, you optimize for learning. In maturity, you optimize for consistency and quality at scale. The goal isn’t to get slower. It’s to get more precise about what speed is in service of. AI is great at accelerating exploration. Helpful in validation. But as products scale, something else matters: When everything is built the same way, it starts to look and feel the same. That’s where craft still matters. Not as decoration — as differentiation. The future isn’t one tool replacing another. It’s knowing what to optimize for at each stage — and having the judgment to shift between them.

15

This podcast episode has been doing the rounds on various design forums on X (sorry), Discord and Slack. The title of the episode is shameless clickbait but also bears unpacking... If you've worked in Product or Product Design in the last decade you'll probably recognise that the 'design process' has been dead for some time — in fact, I'm not sure it ever really existed outside of UX Design courses and textbooks. Design 'process' isn't a linear sausage-making act, it's a tool kit that designers learn to apply contextually to the right problem, at the right time in the right team. How to apply 'just enough' process to a problem is a skill absent in current design education (formal or informal) and takes years to develop. This skill is developed and honed day to day, what is being billed here as 'death' is really the only constant that exists in design and tech — change. AI can help us synthesise, build, measure and learn faster. The feedback loops are shorter and the risk to commit is lower so the discovery loops can be too. The pie chart of daily 'time spent' and stack to ship might look different but the foundational, critical thinking skills remain the same. Disclaimer: I'm a fan of Lenny's work, and the discussion with Jenny was a really interesting snapshot into what bleeding edge design practice looks like at Anthropic. Go listen! https://lnkd.in/e5UdFA-G

4

Design Review & Critique Why are design reviews important? There is a right and wrong was to critique a design idea and which one you employ begins with the setting of expectations. You need to learn HOW to critique. It is not about ripping a thing apart, or “scoring points”. About 20 years ago, I had a VP give me absolutely terrible feedback in a design review. It was one sentence; "This design is shit." Not helpful. Not a very good leader in my book. That was the catalyst for me really thinking about design critique and I eventually put together a presentation that I now share with any team I lead. It was pretty simple. I still use it today. I begin by telling the story of the leader who didn’t know how to lead. I always started this presentation by asking about the image on this post.  Surprisingly some didn’t get it. Design Review & Critique Why are design reviews important? Critiques are meant to improve output, rather than hinder progress. Collaboration and feedback improves your work. The designer who owns the design should own the critique session. Reviews should be conducted early and often in the process. Moderator Have a Moderator to help keep people focused and make sure the presenter has a successful critique. They should make sure the feedback doesn’t veer away from the presenter’s scope; be a time-keeper. The moderator sends anything requiring deeper discussion to a ‘parking lot’. Explain that questions should be held/written down until the end. Presenter The person who owns the design. Explain what you are showing, convey what you have done, and be clear about where you need help. Communicate what stage you are at in the process. Walk through your rationale. Provide context. Accept any feedback graciously & thoughtfully. The commentary is (should be) on the work, NOT you. Take notes during the feedback. Reviewers Those critiquing the design. Ask yourself, ‘How can I help this person improve their work?’ Hold your questions until the presenter is done, unless it is a clarifying question. Write down any feedback you think of. Frame it for understanding and action. Your feedback should be useful. Speak about the design NOT the designer. Avoid the pronoun ‘you’. Understand what they have done. See the opportunities to improve. Be specific about what is working and what is not. Speak from the target-customer-user’s perspective. YOU are not the user. That’s pretty much it. What have you done? How have you handled difficult situations. What feedback do you have for me on this subject? Read more - and view the presentation on my Substack... https://lnkd.in/gwyZRNWp

1

We’ve been here before. When design systems matured, UI stopped being the hard part. Craft didn’t disappear—but it moved. The value shifted from drawing interfaces to deciding why they should exist at all. AI is pushing the next shift. Most of what we call “design work” is already decomposing into patterns: layout, hierarchy, component behavior, presentation logic. That’s fine. It means the surface of products is becoming a solved problem. The real work is upstream: understanding users, shaping intent, defining the decision architecture behind an experience. I’m not nostalgic for pixel craftsmanship. I’m interested in what happens when AI tools make interface production as trivial and standardized as we made UI kits twelve years ago. That’s the moment when the center of gravity in design moves again—this time from the screen to the system. We’re not there yet. But that’s the trajectory

6

2 Comments