Balancing Innovation And Practicality In Engineering

Should you try Google’s famous “20% time” experiment to encourage innovation? We tried this at Duolingo years ago. It didn’t work. It wasn’t enough time for people to start meaningful projects, and very few people took advantage of it because the framework was pretty vague. I knew there had to be other ways to drive innovation at the company. So, here are 3 other initiatives we’ve tried, what we’ve learned from each, and what we're going to try next. 💡 Innovation Awards: Annual recognition for those who move the needle with boundary-pushing projects. The upside: These awards make our commitment to innovation clear, and offer a well-deserved incentive to those who have done remarkable work. The downside: It’s given to individuals, but we want to incentivize team work. What’s more, it’s not necessarily a framework for coming up with the next big thing. 💻 Hackathon: This is a good framework, and lots of companies do it. Everyone (not just engineers) can take two days to collaborate on and present anything that excites them, as long as it advances our mission or addresses a key business need. The upside: Some of our biggest features grew out of hackathon projects, from the Duolingo English Test (born at our first hackathon in 2013) to our avatar builder. The downside: Other than the time/resource constraint, projects rarely align with our current priorities. The ones that take off hit the elusive combo of right time + a problem that no other team could tackle. 💥 Special Projects: Knowing that ideal equation, we started a new program for fostering innovation, playfully dubbed DARPA (Duolingo Advanced Research Project Agency). The idea: anyone can pitch an idea at any time. If they get consensus on it and if it’s not in the purview of another team, a cross-functional group is formed to bring the project to fruition. The most creative work tends to happen when a problem is not in the clear purview of a particular team; this program creates a path for bringing these kinds of interdisciplinary ideas to life. Our Duo and Lily mascot suits (featured often on our social accounts) came from this, as did our Duo plushie and the merch store. (And if this photo doesn't show why we needed to innovate for new suits, I don't know what will!) The biggest challenge: figuring out how to transition ownership of a successful project after the strike team’s work is done. 👀 What’s next? We’re working on a program that proactively identifies big picture, unassigned problems that we haven’t figured out yet and then incentivizes people to create proposals for solving them. How that will work is still to be determined, but we know there is a lot of fertile ground for it to take root. How does your company create an environment of creativity that encourages true innovation? I'm interested to hear what's worked for you, so please feel free to share in the comments! #duolingo #innovation #hackathon #creativity #bigideas

A $12 prototype can make $50,000 of engineering analysis look ridiculous A team of engineers was stuck on a bearing failure analysis for six weeks. Vibration data, FFT analysis, metallurgy reports - they had everything except answers. The client kept asking for root cause and the engineers kept finding more variables to analyze. Temperature gradients, load distributions, contamination levels, manufacturing tolerances. Each analysis created more questions. Then the intern did something that made the engineers feel stupid. She 3D printed a transparent housing and filled it with clear oil so the engineers could actually see what was happening inside the bearing assembly. Took her four hours and $12 in materials. They watched the oil flow patterns and immediately saw the lubrication wasn't reaching the critical contact points. All their sophisticated analysis was based on assuming proper lubrication distribution. Wrong assumption. Six weeks of wasted effort. The visual prototype didn't just solve the problem - it changed how the engineers approach these types of investigations. Now they build crude mockups before diving into analysis rabbit holes. Cardboard, tape, clear plastic, whatever works. Physical models force you to confront your assumptions before you spend weeks analyzing the wrong thing. Sometimes the cheapest prototype teaches you more than the most expensive simulation. #engineering #prototyping #problemsolving

A friend texted me this weekend for some advice. They said their CEO just sent this over and asked how to reply. "If AI tools can really turn junior or average engineers into 10x contributors, why aren’t we doing more? Why does this still feel optional?" I wasn’t surprised. This is the number one question I hear from engineering leaders.  So here’s how I broke it down: 1. Reality today Across teams using tools like Copilot, Cursor, Amazon Q, Gemini Code Assist, Sourcegraph, and others, we’re seeing measurable productivity gains. Studies and data from the Jellyfish platform consistently show: - 30-50% faster throughput for engineers who engage deeply with these tools - Significant reductions in boilerplate work and time-to-implementation - But only one-third of engineers are using them consistently (monthly) It’s a significant gain, but it’s certainly not 10x. And while those gains do happen, they're just not indicative of what you can realistically expect in the common case. We usually see those outsized speedups under the following conditions: - Greenfield teams writing entirely new code - AI-native workflows with heavy automation - Engineers who are deeply enabled, trained, and incentivized to adapt Most orgs today are still operating in legacy environments, where coding new features is a minority share of engineering time, and maintenance, integration and iteration dominate. AI is powerful, but it hasn’t replaced the full software development lifecycle – or at least not yet. 2. What to tell your CEO We shouldn’t push back on the ambition. Instead, we should match it with clarity and action. If your CEO is asking for 10x productivity, answer with the following: “I want that too. Let’s lean in together,” then: - Implement clear telemetry on current AI usage and outcomes - Build a plan to increase engagement and enablement – training, guidance, and accountability - Create a case for investing in deeper experimentation with emerging tools and workflows - Stay tuned in to the state of the art – surface, dig in, investigate and understand the circumstances that made outsized gains possible - And, if needed, ensure there is a willingness to make team and process changes to meet the moment AI in engineering is no longer optional. But we’re still in the very early innings. Just because 10x gains aren’t a reality across the board today doesn’t mean they won’t ever be. Things are moving quickly and as we move out of the IDE to take a more agentic approach, productivity gains will only increase. Success requires more than licenses. It demands leadership, enablement, and serious cultural investment. This is a transformation. Let’s treat it like one.

Are you solving for sustainability or with it? 🌍 There’s a big difference between solving for sustainability and solving with it. When you solve for sustainability, it’s often a bolt-on. The work you do after the “real” work is done. It’s a checkbox. A PR line. An accounting exercise. And honestly, it limits the whole thing—form, function, and financial return. But when you solve with sustainability—when it’s in the room with you alongside the other big drivers like user experience, technical feasibility, financial ROI, and regulatory realities—it becomes part of the innovation DNA. This is what we mean by shifting left, as outlined in The Insider's Guide to Innovation at Microsoft. Responsible Innovation practices—like privacy, accessibility, and sustainability—deliver the most value when they’re baked in from the start. When you shift sustainability left, it stops being a constraint and starts being a catalyst. It allows you to aim for positive—not just less harm, but more value. For business. For people. For the planet. Solving with sustainability doesn’t mean sacrificing profit—it means redesigning the value chain. And when you do that, the returns speak for themselves. Don’t bolt it on. Build it in. Happy Earth Day. Let’s keep aiming for positive—and building a future that benefits us all. #innovation #sustainability #leadership

𝐋𝐞𝐬𝐬𝐨𝐧𝐬 𝐟𝐫𝐨𝐦 𝐇𝐢𝐬𝐭𝐨𝐫𝐲: 𝐍𝐚𝐯𝐢𝐠𝐚𝐭𝐢𝐧𝐠 𝐭𝐡𝐞 𝐀𝐈 𝐂𝐨𝐝𝐢𝐧𝐠 𝐑𝐞𝐯𝐨𝐥𝐮𝐭𝐢𝐨𝐧. As history consistently teaches us, every major production boom – from the Gutenberg Press to the Ford Assembly Line – inevitably brings forth new challenges alongside its transformative opportunities. Based on research from Georgian, in partnership with FirstMark (a benchmarking report on how AI is being adopted by R&D and GTM teams surveying 308 R&D leaders), my FirstMark colleague David Waltcher and I discuss in this episode how we're now witnessing this very phenomenon unfold with the 𝐀𝐈 𝐜𝐨𝐝𝐢𝐧𝐠 𝐰𝐚𝐯𝐞, which is rapidly redefining engineering at an unprecedented pace. In just 24 months, companies like Cursor, Windsurf, Replit, Lovable and products like GitHub Copilot are seeing explosive growth, with millions of developers already leveraging AI to write code. We're already seeing 𝐬𝐢𝐠𝐧𝐢𝐟𝐢𝐜𝐚𝐧𝐭 𝐠𝐚𝐢𝐧𝐬: * 30-50% faster throughput in engineering processes. * 12% increase in PR merges. * 17% more time on roadmap vs. maintenance. However, the current AI coding revolution is no different from past industrial shifts in that it 𝐚𝐥𝐬𝐨 𝐢𝐧𝐭𝐫𝐨𝐝𝐮𝐜𝐞𝐬 𝐧𝐞𝐰 𝐜𝐨𝐦𝐩𝐥𝐞𝐱𝐢𝐭𝐢𝐞𝐬. We're now navigating: * Increased 𝐝𝐞𝐛𝐮𝐠𝐠𝐢𝐧𝐠 time and rising 𝐬𝐞𝐜𝐮𝐫𝐢𝐭𝐲 𝐯𝐮𝐥𝐧𝐞𝐫𝐚𝐛𝐢𝐥𝐢𝐭𝐢𝐞𝐬. * Performance issues and 𝐨𝐯𝐞𝐫𝐰𝐡𝐞𝐥𝐦𝐞𝐝 𝐐𝐀 𝐩𝐫𝐨𝐜𝐞𝐬𝐬𝐞𝐬. * 𝐀 𝐟𝐚𝐬𝐜𝐢𝐧𝐚𝐭𝐢𝐧𝐠 𝐬𝐡𝐢𝐟𝐭 where experienced engineers are becoming professional code reviewers. This parallel with past industrial shifts presents immense opportunities for CTOs, engineers, founders, and investors: * 𝐂𝐓𝐎𝐬 face critical decisions on talent, architecture, and security, much like leaders during previous technological paradigm shifts. * 𝐇𝐢𝐫𝐢𝐧𝐠 𝐢𝐬 𝐞𝐯𝐨𝐥𝐯𝐢𝐧𝐠, now valuing great editors, reviewers, and prompt engineers – a new kind of "specialist" emerging from a new era of production. * 𝐍𝐞𝐰 𝐬𝐨𝐥𝐮𝐭𝐢𝐨𝐧𝐬 are emerging in automated security, reasoning-based QA, and agentic code checks, mirroring the new industries that sprung up to solve problems created by past revolutions. The future of engineering is rapidly evolving. What historical parallels do you see in today's AI transformation, and how are you preparing for what's next? #AI #Engineering #FutureOfWork #TechInnovation #CTO #SoftwareDevelopment #FirstMark https://lnkd.in/e5sGaaH7

To students of industrial design: In the world of design, one question often overlooked is the intersection of creativity and practicality: “At what stage should one consider the manufacturing cost?” This query, posed by a member of Young Designers India (YDI), highlights a common oversight in design education. Business and manufacturing considerations are rarely covered in depth in design schools, yet they are crucial for success in the professional realm. In my practice as an industrial design consultant, I begin each project by understanding the client’s business goals. This includes batch size, budget, cost of goods, etc. Such questions do not stifle creativity; rather, they provide a framework for exploration. Reflect on this: industrial designers often aim to develop products for mass production to achieve economies of scale. However, initial versions or beta products usually do not meet this criterion. Clients seek market feedback to refine their offerings, which means your design must be feasible and scalable from the outset. Consider the entire journey of your product, not just the end result. Think about materials, manufacturing processes, packaging, and logistics. This holistic view helps you understand the costs that will accrue down the supply chain. During the conceptual phase, let your ideas run wild. But as you refine your concepts, measure them against the constraints of cost and manufacturing feasibility. This balance preserves innovation while ensuring practicality. Great design solves user problems but must also be viable in the real world. Balance creativity with practical constraints. Ask questions, understand the scope, and consider every aspect of the journey from concept to consumer. Joining a community like Young Designers India (YDI) can provide invaluable insights and support. Engaging with professionals and peers can deepen your understanding of these crucial aspects of design. If you’re interested in joining YDI, find the link in the comments below. #IndustrialDesign #ProductDesign #Manufacturing #DesignProcess #YoungDesignersIndia For further discussion or guidance, please feel free to DM.

Boeing engineers once filled an entire airplane with sacks of potatoes just to test the in-flight Wi-Fi. It happened around 2012 when the company needed a reliable way to fine-tune their wireless signal strength for passengers. They needed to simulate a plane full of people, but having human testers sit motionless for days on end was not a practical solution. They discovered that potatoes, due to their specific water content and chemistry, absorb and reflect radio wave signals in a way that is remarkably similar to the human body. So, they loaded a plane with thousands of pounds of potatoes, placing a large sack in every single seat to mimic a full flight. This allowed them to systematically map the signal strength throughout the cabin, identifying weak spots and dead zones that needed to be fixed. The method, while unusual, was a clever and effective piece of engineering that helped ensure a better connection for travelers. This potato-based testing provided the data needed to optimize the placement of Wi-Fi routers and signal boosters on their aircraft. Sources: Journal of Food Science, Phys org

We have one rule in my org for evaluating tech projects called "The One Miracle Rule" 🔮 When assessing any complex initiative, map out every critical element: technical challenges, resource needs, timeline constraints, team dependencies, and organizational changes. Here's the rule: You get ONE "miracle" - ONE major unknown you'll need to solve along the way. That's your innovation space. But if you need multiple miracles (like inventing novel ML features/signals AND building unprecedented infrastructure AND collaborating with two or more separate orgs), it's time to pivot: "Neat idea, maybe when it's a one-miracle project..." Why this works: Innovation thrives on pushing boundaries, but execution demands pragmatism. One miracle? That's ambitious yet achievable. Multiple miracles? That's where I've seen too many projects spiral into missed deadlines and burned-out teams. The most successful projects I've led weren't necessarily the most ambitious - they were the ones that found that sweet spot between innovation and realistic execution paths. Interesting twist: In today's LLM era, many previous "miracles" in NLP have become "just" difficult engineering challenges. But productionizing these capabilities at scale? That might still count as your one miracle, depending on your requirements. I'm aiming for those LLM productionization to become "business as usual" in my team.

"The value of a prototype is in the insight it imparts, not the code" Prototyping lets us fail fast and cheap, or get the data to make a concrete decision on direction. It helps answer the question, "What happens if we try this?". Most significantly, prototyping provides us with the guardrails to safely and productively fail. Prototyping is the right tool if you have an idea to validate, a clear path to get feedback on, or a proposal requiring further data. It provides crucial insights to move forward. By creating a rough version of a feature or system you've been considering, you gain the flexibility to either discard the idea or fully commit to it. It's a skill that assists product and engineering teams in making pivotal business decisions. Whether it's a website, mobile app, or landing page, no matter what product you're working on, it's always essential to verify your design decisions before shipping them to the end-users. Some development teams delay the validation stage until they have a solution that is almost complete. But that's an extremely risky strategy. As we all know, the later we come across the problem, the more costly it becomes to fix it. Luckily, no matter what point you are in the design process, it is still possible to build and test a concrete image of your concept—a prototype. Consider an architect tasked with designing a grand building. Before laying the first stone, the architect crafts a miniature scale model, allowing them to visualize the end result, understand the project's complexities, and present their ideas convincingly to others. However, this model is far from being the final product; it's a means to an end. This principle applies just as aptly in the world of software development. A software prototype—whether it's a low-fidelity wireframe, a high-fidelity interactive model, or a simplified mock-up of a more complex system—is much like the architect's scale model. It's a visual, often interactive, model of the software that provides developers, stakeholders, and users with an early glimpse into the software's workings, long before the final product is ready. The prototype isn't about the code per se; the code is merely a tool used to create it. Instead, it is about gathering valuable insights, comprehending user needs, identifying functional requirements, validating technical feasibility, and discovering potential stumbling blocks that might arise during full-scale development. The prototype's strength lies in its capacity to provide these insights without necessitating a significant investment of time or resources. I'm a big fan of using prototypes in our work at Google. Their value is often high. Wrapping up... The aim of prototyping is not the prototype itself or its immediate output but the knowledge that comes from it. I wrote more on this topic in https://lnkd.in/gEEGFwJp #softwareengineering #programming #ux #design

Throwing AI tools at your team without a plan is like giving them a Ferrari without driving lessons. AI only drives impact if your workforce knows how to use it effectively. After: 1-defining objectives 2-assessing readiness 3-piloting use cases with a tiger team Step 4 is about empowering the broader team to leverage AI confidently. Boston Consulting Group (BCG) research and Gilbert’s Behavior Engineering Model show that high-impact AI adoption is 80% about people, 20% about tech. Here’s how to make that happen: 1️⃣ Environmental Supports: Build the Framework for Success -Clear Guidance: Define AI’s role in specific tasks. If a tool like Momentum.io automates data entry, outline how it frees up time for strategic activities. -Accessible Tools: Ensure AI tools are easy to use and well-integrated. For tools like ChatGPT create a prompt library so employees don’t have to start from scratch. -Recognition: Acknowledge team members who make measurable improvements with AI, like reducing response times or boosting engagement. Recognition fuels adoption. 2️⃣ Empower with Tiger Team Champions -Use Tiger/Pilot Team Champions: Leverage your pilot team members as champions who share workflows and real-world results. Their successes give others confidence and practical insights. -Role-Specific Training: Focus on high-impact skills for each role. Sales might use prompts for lead scoring, while support teams focus on customer inquiries. Keep it relevant and simple. -Match Tools to Skill Levels: For non-technical roles, choose tools with low-code interfaces or embedded automation. Keep adoption smooth by aligning with current abilities. 3️⃣ Continuous Feedback and Real-Time Learning -Pilot Insights: Apply findings from the pilot phase to refine processes and address any gaps. Updates based on tiger team feedback benefit the entire workforce. -Knowledge Hub: Create an evolving resource library with top prompts, troubleshooting guides, and FAQs. Let it grow as employees share tips and adjustments. -Peer Learning: Champions from the tiger team can host peer-led sessions to show AI’s real impact, making it more approachable. 4️⃣ Just in Time Enablement -On-Demand Help Channels: Offer immediate support options, like a Slack channel or help desk, to address issues as they arise. -Use AI to enable AI: Create customGPT that are task or job specific to lighten workload or learning brain load. Leverage NotebookLLM. -Troubleshooting Guide: Provide a quick-reference guide for common AI issues, empowering employees to solve small challenges independently. AI’s true power lies in your team’s ability to use it well. Step 4 is about support, practical training, and peer learning led by tiger team champions. By building confidence and competence, you’re creating an AI-enabled workforce ready to drive real impact. Step 5 coming next ;) Ps my next podcast guest, we talk about what happens when AI does a lot of what humans used to do… Stay tuned.