Devpost’s Post

When AI capabilities are getting better every week, the gap between “I have an idea” and “let’s plan this for Q3” is where good ideas go to die. That’s why we run hackathons twice a year, at every company offsite. Everyone participates, not just engineers. Anyone can pitch. Teams form organically. There’s no pressure to ship production-ready code. And yet, some of our most-used features started exactly this way. One engineer used a two-day hackathon to build an AI-powered tool that semantically links PRs to related issues. A few weeks later, it was in production. By organizing company-wide hackathons often enough, we’re intentionally building a culture where experimentation is cheap and fast. Read more about how we run our hackathons, and why you should run them too: https://swarmia.co/3O5CtQx

We’re excited to share that EventSentinel.ai has been invited to participate in an upcoming global hackathon. Hackathons are more than just innovation sprints, they’re real-world testing environments where ideas meet execution, and where technologies are pushed under pressure. This initiative builds on experience from 400+ hackathons in collaboration with leading organizations such as OpenAI, Anthropic, ElevenLabs, Google, Harvard University, Replit, and Project Europe, bringing together a global community of builders. What makes this particularly interesting for us is the opportunity to engage directly with developers, DevOps, and infrastructure engineers building in real time, while gaining valuable insights into how tools perform in dynamic, high-intensity environments. At EventSentinel.ai, we see these moments as critical: not just for innovation, but for understanding how systems behave under real-world conditions, where coordination, resilience, and adaptability truly matter. #Hackathon #Innovation #AI #TechEvents #DeveloperCommunity #Resilience #Infrastructure #EventSentinel #AIforGood

Hackathons are so much fun! My team has participated in hackathons to discover new ideas that we can experiment with quickly. These team events help to answer key questions for innovations like: what is going to make this idea fail? How can I find new ways around roadblocks? Have I discovered something nobody else has seen? Is there a way I can nurture this idea and make it grow big? Hackathons aren't just about quickly solving a problem and doing a fancy demo. It's using all the tools at your finger tips to find more ways to succeed; even if it means just re-using an existing tool for a new idea. Recently, my team performed an AI focused Hackathon within our team. Our goal was to use AI tools and knowledge to make our day jobs easier. This was Customer Obsession where we are the customers and the product is our own day-to-day activities. This turned out to be absolutely amazing! Developers found many ways to deliver code safely and securely with AI agents, enhance managing deployments and monitoring production launches with proactive analysis, and even found ways to streamline our organizational calendars. All with AI tools and technologies within 2.5 days! AI Adoption isn't about replacing roles and jobs. Its about reinventing how we do our work so that we can focus on solving more complex problems and challenging ourselves to "solve world hunger". It allows us to really Think Big by freeing up resources, investments, and time to maximize return on innovation.

The recent CoinDCX Internal Hackathon was genuinely eye-opening, it showed me what happens when you start thinking of AI as your companion. The velocity and productivity it unlocks will change how you think about building entirely. We had 55 teams across functions building non-stop for 24 hours, focused on growth, efficiency, and trust. While the participants' excitement and the room's energy were impressive, the speed of creation impressed me the most. We had planned resources carefully. But teams ended up consuming AI credits 2x faster than expected. The myth got busted! In this AI era, exploration and critical thinking are the bottlenecks. Teams weren't stuck on "can we build this?" They were asking, "Should we build this… or something better?" And that changed the role of the hackathon completely. It becomes less about prototyping, and more about questioning the problems worth solving. - what consumer pain points are we addressing - what impact are we creating - whether we're truly delivering value through what we build. I saw ideas move from concept to working demo in hours. And more importantly, several of them are now moving into performance testing and potential product integration. We're moving into a world where software creation is no longer a bottleneck for business growth. Distribution, brand trust, and customer experience become even more critical moats, alongside the value of what you're actually delivering through your product. Hackathons, powered by AI, are now the fastest way to stress-test all of that, and find your next real advantage.

You should be running internal hackathons at your company. Most teams are massively underestimating how buildable things are right now and are stuck doing things the way they always have. But when you give your team a few hours and force them to build… Something interesting happens: They struggle… hit a few dead ends… ask the right questions… then something clicks. And once it clicks? They stop thinking “I can’t build this” and start seeing opportunities for improvement everywhere in their own work. The shift is no longer technical. It’s psychological. Here’s how I’d do it: - Spin up something simple (n8n, Zapier, Claude Code, etc.) - Give everyone API access (OpenAI / Claude + core tools) - Tell them to use AI as a tutor - Block ~3–4 hours total - Let people work solo or in pairs - Have one go-to person for questions - Force a demo day (this is important! they must showcase their work.) - Set the expectation: V1 just has to work You’re not trying to build anything perfect. You’re just giving people structured time to "play" with AI. That usually doesn’t happen until it’s made mandatory with everything set up and a push for everyone to showcase what they built. You’re breaking the “I’m too busy / I can’t build this” barrier. That’s the real blocker for most people. Once you push through it once, you start seeing automation opportunities everywhere.

Most companies run internal hackathons completely wrong. They treat them like high-pressure feature factories. The unspoken expectation is always: build our product roadmap, work through lunch, and ship something production-ready by 5 PM. No wonder they fail to drive real culture shifts. A couple of months ago, we ran an all-company AI hackathon to drive internal adoption, but we set a very different baseline: no pressure. This wasn't a sneaky way to clear our backlog. The goal was purely to get everyone - Sales, CSM, Marketing - excited about AI, sharing knowledge, and building something for fun (or just automating the most annoying part of their day). But you can't just tell non-technical teams to "build with AI" and walk away. To remove the friction, we structured it intentionally: 💡 The Buddy System: We specifically paired one engineer with one non-engineer to form teams of two. Ops brought the domain expertise and the daily pain points; Engineering brought the technical execution. 💡 The Translation Layer: I put together an "AI Development for Non-Engineers" presentation and step-by-step onboarding docs so nobody was staring at a blank screen. 💡 The Infrastructure: I vibecoded a custom internal app to manage the event, track submissions, and handle the voting. 💡 The Support: We had Eng Leads acting as floating mentors to unblock teams on the fly, plus a panel of judges to keep it fun. We gave them half a day to just play around. The result? 28 shipped projects. Some were purely for laughs. But several of them were so practical that they instantly became internal tools our team uses every day in operations. Hackathons aren't a magic bullet for AI adoption. But when you remove the pressure and focus on curiosity, people actually want to build.

𝗕𝗿𝗮𝗶𝗻𝘀𝘁𝗼𝗿𝗺𝗶𝗻𝗴 𝗵𝗲𝗹𝗽𝗲𝗱 𝗺𝗲 𝘄𝗶𝗻 𝟴 𝗵𝗮𝗰𝗸𝗮𝘁𝗵𝗼𝗻𝘀. At Hack the North, Satyam Singh and I spent more than half the time brainstorming, exploring, and prototyping. We finished our project in under 8 hours - and still won. One of the most frequent questions I get in DMs is how to win at hackathons. After 8 wins (including UofTHacks and Anthropic), here's what I learned: 𝗮 𝗴𝗼𝗼𝗱 𝗶𝗱𝗲𝗮 𝗯𝗲𝗮𝘁𝘀 𝗳𝗮𝘀𝘁 𝗲𝘅𝗲𝗰𝘂𝘁𝗶𝗼𝗻 𝗲𝘃𝗲𝗿𝘆 𝘁𝗶𝗺𝗲. Most teams lose before they write a single line of code. They pick the first idea that sounds cool, sprint for 24 hours, and wonder why the judges weren't impressed. The winning teams spend time figuring out what to build - and why it matters. Here's the framework I use! 𝟭. 𝗦𝘁𝗮𝗿𝘁 𝘄𝗶𝘁𝗵 𝘁𝗵𝗲 𝗷𝘂𝗱𝗴𝗲𝘀, 𝗻𝗼𝘁 𝘁𝗵𝗲 𝘁𝗲𝗰𝗵 Before ideating, ask: what would make a judge stop and lean in during a 3-minute demo? 𝟮. 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗲 𝘄𝗶𝗱𝗲𝗹𝘆, 𝘁𝗵𝗲𝗻 𝗳𝗶𝗹𝘁𝗲𝗿 𝗵𝗮𝗿𝗱 Spend 30–60 mins throwing out anything. No filtering yet. Then cut with one question: can we demo this compellingly in 3 minutes? 𝟯. 𝗣𝗶𝗰𝗸 𝘁𝗵𝗲 𝗯𝗲𝘀𝘁 𝘀𝘁𝗼𝗿𝘆 The clearest problem → solution → impact narrative will win over a muddled narrative. 𝟰. 𝗦𝗰𝗼𝗽𝗲 𝗿𝘂𝘁𝗵𝗹𝗲𝘀𝘀𝗹𝘆 𝗯𝗲𝗳𝗼𝗿𝗲 𝘆𝗼𝘂 𝗯𝘂𝗶𝗹𝗱 Map out exactly what you're building and what you're cutting. A polished MVP beats a half-finished ambitious project every time. It's definitely uncomfortable watching other teams code while you're still at the whiteboard. But do it anyway - the hours you spend on the right idea save you from 24 hours building the wrong one. With AI, coding is no longer the bottleneck - ideas are. 👇 Got questions about the brainstorming process? Drop them below :)

Working in tech feels like A LOT right now (a recent post from Elena Verna resonated with many!) On one hand, everyone's beyond excited by what's become possible with AI. On the other, there's almost a paralysis that comes from the crazy speed, and a constant pressure to make decisions that will move things forward, not just the ones that feel exciting in the moment. Something like a hackathon is great for exactly this reason: it shows what's possible, it proves you don't need weeks of thinking & tinkering to build something that works, and there's something about learning by doing that no training deck can replicate. But doing and understanding are not the same thing. You can leave a hackathon energized and still not know how to take the next step on your own. That's the gap enablement is meant to close. Solid enablement starts with understanding your audience and empathizing with where they actually are, not where you'd like them to be. It means providing the building blocks that allow someone to connect new knowledge to what they already know, and advancing from there step by step, with each stage explained along the way. Yes, it takes more time than a hackathon but that's what leads to learning. Leaders and everyone else navigating this: given how fast things keep shifting, how do you approach your own learning and how do you enable others to learn what you know?

I’m thrilled to be joining Ethan S. Sumner for the first-ever Platform Engineering Leaders meetup! We’ve spent 20 years trying to make software delivery "deterministic", binary, predictable, and controlled. But as we move into the era of AI, we’re facing a new challenge: The Probabilistic Age. I’ll be diving into why our traditional "Gatekeeping" needs to evolve into "Continuous Observation" and how leaders can build the control planes necessary to manage AI without losing the speed or financial oversight their organizations demand. Looking forward to seeing a great crowd there. Sign up below! 🚀 #PlatformEngineering #Leadership #AI #DevOps LaunchDarkly