Workshop AI ads

🏆 Built a World Cup 2026 Match Calendar because I couldn't find one that worked. I'm planning which matches to buy tickets for, and every schedule out there is either cluttered, hard to filter, or missing a proper bracket view to understand the knockout matchups easily. So I built what I needed with Memex. 🔍 Key features: • Complete 104-match schedule with filtering by date, venue, and stage • Interactive knockout bracket showing how matches flow into each other (in both directions) • Details on all 16 venues across USA, Canada & Mexico (including public transportation and driving directions) • Save favorites and share specific matches If you're looking for tickets, planning a trip, or just want to get smarter on the new expanded structure, hopefully this helps! The link it's in the comments. I'll update it with specific team details after the draw on December 5th. And if you need something similar built for your own use case, that's kind of the point of what we're building at Memex. What matches are you targeting? I have tickets for a couple of already, but still trying to find for the quarter finals match in Boston (M97) #WorldCup2026

I didn’t expect a side experiment to change how I think about the future of Memex. We’ve been testing how to support local models as they become viable for agentic coding. So I used Memex itself to run the tests: Mojo/Modular, llama.cpp, LM Studio, ollama … All inside our own tool. One prompt stood out. It asked Memex to build a Tauri app that embeds llama.cpp directly. The first version failed. I tweaked the prompt. The next one worked: A desktop app that chats with local models using my GPU. It’s not ready for release. But it’s an example of a prototype that we use as a part of our product-development lifecycle. And it feels like a glimpse of what’s next: Private, fast, local AI workflows that never touch the cloud. That’s where agentic coding is heading. And it’s closer than most think. If you want to see the exact prompt that worked, DM me and let’s connect. Happy to share it.

Want your vibe coded creations to feel like they belong in your product—not just a generic UI? That’s where component libraries come in. They’re curated examples you can pass to AI to ground every prompt in your app’s design. - If you’re starting on a fresh app, try something like 21st.dev - Or if your company has its own design system, use it. Whatever your scenario, Memex supports it. Connect to Figma via MCP to bring design components in, or reference the source code of components directly like Dom does in this video [1] Consistent components mean every generated screen speaks your language and stays on brand. What has your mileage been getting AI to produce consistent design outputs?

Every time you say “Hi” to an AI, you might be sending thousands of hidden words under the hood. Most people assume AI cost equals prompt size. It doesn’t. The real drivers are (1) context size, (2) model type, and (3) how fast your conversation compounds over time. When you send a message to an AI service, you’re not just sending one line. You’re resubmitting: – The system prompt (which can be tens of thousands of words long) – Your project’s context (e.g. a memory or code-base index) – The entire conversation history – And finally, your new message That invisible load is what drives real costs. The upside? You can control most of it. Once you understand what’s happening under the hood, you can design your workflow to spend less and build faster. After testing hundreds of projects in Memex, I found four techniques that make the biggest difference. They cut costs, speed up iteration, and and improve outcomes. If you’re using Memex or any other agentic coding / vibe coding tool, watch this breakdown. It’ll show you exactly how AI costs work, and how to make every token count. Watch the full video on YouTube (link in comments), and consider joining 2,500 other people that subscribe to our content there.

🚀 If Memex existed back when I was at Stanford GSB… …I think half my classmates would’ve launched something before graduation. At any given lunch table, you’d hear the same three ambitions: 👉 “I want to start a company.” 👉 “I want to be a PM.” 👉 “I want to build my own business someday.” The ideas were everywhere — marketplaces, AI tools, fintech apps, new D2C brands — but turning those ideas into something real always hit the same wall: “How do I build it?” Technical execution was the bottleneck. Whether you wanted to: 💡 build an MVP to test an idea, 📊 run a financial or growth model, 📈 analyze marketing campaign data, 🔍 research and visualize a new market, ⚙️ or create a small internal tool to automate a process, you needed someone technical to help. If Memex had existed then, you could’ve just said: “Spin up a prototype that connects to Stripe, stores user data, and generates revenue dashboards. Run market-sizing models in Python and financial projections in Excel. Analyze early campaign results with SQL and visualize them in a simple web app.” Memex would: ✅ Write and run the code itself (across languages and tools) ✅ Fix issues and iterate automatically ✅ Combine local and cloud workflows (from your desktop to hosted environments) ✅ Deploy everything to a shareable prototype or dashboard In a few hours, what once needed a full dev team could be live; ready to test with users, investors, or classmates. That’s the 0→1 moment where most ideas stall. Memex gives you a way through it. For founders, PMs, and builders, it’s the difference between having an idea and proving it. We built Memex for the people stuck between idea and execution. The ones who can describe what they need, but shouldn’t have to code it from scratch.

We just crossed 10,000 users at Memex. And now I get some variation of this message almost every day. When I ask users “why do you prefer Memex to other tools”, the answers follow the same themes: - It was successful for them where other tools failed. - It doesn’t limit them to a single tech stack … that tool you want to use for your build? You can use it with Memex. - It’s great for technical people … even better for non-technical people. - It has the best coding agent quality. When designing an AI gent, there are three dimensions: - quality - speed - cost You can’t have all three. At Memex we’ve always prioritized quality. The rationale is simple: it doesn’t matter if a coding agent is faster or cheaper if it doesn’t produce outcomes. Your time is your most important asset. And producing real outcomes is the way we can best respect it. That’s why we’re all-in on agent quality. And vibe coders are starting to take note.