Overview
Dashdot combines startup speed with enterprise security to build AI-driven digital products for customers and themselves. The studio serves startups, SMEs, and enterprises.
Key people
In the news
- Burgers? Smashed. Furniture? Moved. Office? Warmed. Thank you to everyone who showed up. It was great to see our office filled with not only colleagues, but clients, partners, friends and family too. Shoutout to Webclew's former intern Klaas who went from writing code to grilling burgers. You guys smashed it 😉 When you walk around 't Eilandje, look for our stickers on the door and come say hi!
- Vibecoding vs. Agentic Engineering. Our 2 cents. Vibe coding starts cheap. You describe what you want, something appears, you're looking at a working screen the same afternoon. Nothing competes with that for speed to first output. After a while the line turns, and it turns hard. Past a certain point, every new feature costs 3 to 10 times more than the same feature on a properly set up codebase. Prompting tax, then maintenance tax, then regressions you only hear about because a real user hit one. Agentic engineering is the opposite
- Ahh, September... Temperatures are dropping, days are getting shorter and commutes are getting busier again. Technically still summer, just slightly worse in every way. Except at Dashdot HQ - where we brightened up our office ánd our days with not one, but TWO new hires! 🥳 Talk about a mood booster 🤩 Welcome Tycho Verstraete and Pieter Van Overbeke!
- Our favourite moments during the Q&A last event: ► Our CTO described his job: read the spec, correct it, read the plan, correct it, press implement, then read the pull request. That's becoming a typical day. You delegate the typing. You don't delegate the thinking. Turns out the thinking was always the job, we just used to get little breaks from it while our hands caught up. ⌨️ ► Someone asked how we handle the ten minutes of waiting while an agent works. Sometimes we make a coffee, and that's fine. Mostly we start the next
- One of the skills we use has exactly one job: make the AI talk like a caveman. It strips out conjugations, filler, connecting words, anything that isn't load-bearing. Purely to use fewer tokens. It reads ridiculous. And the output quality is the same. That's the part that bothers me. Everything we once assumed was helping, the polite phrasing, the full sentences, the careful setup, was decoration. The model never needed it. We did, because humans find it weird to bark at a machine. (And we’re still quite polite, you never
- Some of the people shipping production features at Dashdot used to be designers. Well, they still are. But now they’re shipping features too. Not prototypes. Real features, that real users touch. A year ago I would have called that reckless. 😅 The thing that changed isn't that they suddenly learned to code. It's what we built around them. Because the real challenge wasn’t getting AI to write code. It was knowing when to trust it. AI gives you output that sounds confident. Telling that apart from output that’s actually correct is
- If you are a startup founder in our network looking to scale, this is your chance! Applications for the next wave are officially open! 📅 Info session: 08.09.2026 → https://lnkd.in/eVZX5CnS ⏳ Applications close: 14.09.2026 at midnight → https://lnkd.in/eFe9RMM3 #StartitKBC #Startups #Innovation #Entrepreneurship #Founders
- 50 people on a Monday evening? Damn right. And they were there with one goal. 🫡 To push the boundaries of what's possible with agentic coding. So we opened up and showed what's under our hood. To share how we try to tackle the hard parts, and te learn from others how they're doing it. Thank you Jonathan and Lorenzo from HappyClient to show us how you can structure your company to become an AI first organisation. From product to operations, you challenged our assumptions about what's possible. 🤩 A huge thank you to everyone who
Something wrong or missing? Send an update. Fixed within 24 hours.