Kaspar Loog

Quick lesson that I learned building Heimdahl along the way: Talk to people. please! Just chat, learn what their problem actually is and what their problem IS NOT!!! They are happy to give you all the details. It's not that hard. If you are developer talk to fellow devs that use some tool X or Y or Z. And iterate on it. that's it. No magic. No rocket science. No pitches and no disappointments. It just works. It's 80% communication and 20% execution. #bootstraplife #building #productdev

10

Last week I saw a post on my timeline asking: Why can children play games for hours but can’t focus on homework for 5 minutes? The explanation was simple. In a game, something happens immediately. You click. You move. You try. The system responds. Homework? You finish it… and nothing happens. No feedback loop. And it made me realize something about myself. I generally struggle with focus. My brain jumps. New ideas distract me. I rarely sit still on one thing for hours. Except when I’m “vibe coding.” This weekend it happened again. I started with the idea of quickly exploring a concept… and looked up hours later with a working prototype. No pressure. No roadmap. Just: Prompt → result Tweak → result Break → fix → result Iteration → improvement It feels like a game. Every prompt is a move. Every output is feedback. Every error is a puzzle. Winning the game? Getting it to work. That instant feedback loop completely changes the experience. It doesn’t feel like work. It feels like momentum. Maybe the issue isn’t focus. Maybe it’s feedback. When the loop is tight, motivation increases. When progress is visible, effort feels lighter. When experimentation is safe, curiosity takes over. AI isn’t just powerful because it can generate code. It’s powerful because it turns building into a playable system. And that might change more about how we work than we realize.

12

5 Comments

Can launching a product be compared to training for a marathon? We think so. Because great product thinking starts far beyond the desk. Meet Tasha Kobzarenko, Product Marketing Owner at Storware. Her passion for running became more than a habit—it turned into a mindset that fuels her work. Every day, steps turn into kilometers. Awareness sharpens — from internal focus to external impact. And the same applies to building great products: 🔹Breaking down big goals into small milestones 🔹Staying tuned in to users 🔹Adapting to every shift in the market Watch how Tasha brings the same drive, discipline, and purpose to the road —and to the Storware team. Wideo made by Szymon Fabrowicz! Thank You Szymon! #productmarketing #backupmonster #employeeadvocacy

15

8 Comments

6% more customers, 14% conversion improvement, and it's not a store for everyday-use products. I'm talking about a used car listings platform. The result of increased transparency in vehicle sales for Autoplac.pl, with reliable damage reports on car sales listing pages. The challenge? 🔹 Aggregating real-time car data from multiple sources 🔹 Handling complex data synchronization across platforms 🔹 Delivering high-speed search functionality Link to the whole case study in the comments. #APIIntegration #AdvancedAPIs #APIStrategy #IntegrationSolutions #Netkodo

2

3 Comments

I agree with Satya Nadella. A large chunk of the value of AI will come from diffusion. That is why it’s important to upskill as many people as possible in Estonia 🇪🇪 and across Europe 🇪🇺

1

When we started talking about building Taito, we realized pretty quickly we were missing a key piece. I could handle engineering. Kristo could handle business. But design? I completely lack the eye for it. And we both felt like how a product looks and behaves is so crucial, especially for something people use every day, that we couldn't just "figure it out later." Either we'd need to hire a designer immediately, or get someone into the founding team who actually had that skill. Then Mikko became available. Full-stack developer. Amazing design eye. Great at managing projects. Exactly what we needed. When we started discussing with him, everything clicked. The three of us covered the bases: engineering, business, and design; plus we all share the vision of empowering engineers. That's actually how we ended up building without product managers. Not because we're ideologically opposed to PMs. I've worked with really good ones. But with Mikko’s skillset, and my own coming from Solutions Engineering and Customer Success, we could stay close to customers without adding layers in between. Looking back, getting the founding team right was probably the most important decision we made early on. You can pivot on features. You can change your target. But who you're building with? That's harder to undo.

103

5 Comments

Excited to share that datakyte are now officially listed on Built in Baltics We’re building a new AI-native distribution layer for travel that turns videos, creator content, social posts, messaging apps, websites, and AI agents into direct booking channels. The #Baltics are becoming one of the most exciting places in Europe for ambitious technology companies, and we’re proud to be building from our home in Riga, Latvia 🇱🇻 alongside so many talented founders, builders, and operators. Big things ahead. 🌍 https://lnkd.in/dWJDH7Pk

17

3 Comments

What really makes agentic software unique from traditional software? And is it all-or-nothing? The essential concept is already there in the name: agency. You can control the degree of agency with tool design and human-in-the-loop patterns.

22

6 Comments

Dev teams often mistake surface polish for strategic clarity. ​​A sleek UI might win design awards, but if it’s built on shaky assumptions, it’s just an expensive mirage. If your idea is going to support real users, scale with your business, and withstand inevitable pivots, it needs more than aesthetics. It needs structure. It needs a blueprint. If a project kicks off without a validated plan that accounts for business goals, user needs, tech constraints, and delivery realities, things can go sideways fast. A presentation slide is made to sell an idea. A blueprint is made to build it. ❌ Slides don’t clarify scope. ❌ They don’t define trade-offs. ❌ They don’t align delivery teams with decision-makers. A blueprint does all of that and ensures that every hour spent building moves you toward something that works, lasts, and supports the business. At Mavric, we believe innovation should serve your business, not the other way around. Our role is to bring strategic thinking and technical expertise to your vision, staying grounded in your goals every step of the way. Strategy beats sizzle in every successful software build. Build it right, or don’t build it at all. Are you building from a blueprint or a pitch deck? #productstrategy #softwaredevelopment #digitaltransformation #buildbetter

30

A new year usually starts with strategy discussions, priorities, and one uncomfortable question: “What is our actual plan for using AI this year?” Most organizations already experimented in 2024–2025. Tools were tested. Pilots launched. Some value captured. But very few teams start the year with a clear, shared answer to:  • where AI should be used,  • where it should not be used,  • and how it becomes part of everyday work and processes rather than scattered experiments. That gap between intention and execution is exactly what I want to address. Starting January 2026, I’m running a 𝗳𝗿𝗲𝗲 𝟴-𝘄𝗲𝗲𝗸 𝗽𝗿𝗮𝗰𝘁𝗶𝗰𝗮𝗹 𝘄𝗲𝗯𝗶𝗻𝗮𝗿 𝘀𝗲𝗿𝗶𝗲𝘀 on AI implementation for mid-sized and large organizations — designed to help leadership teams begin the year with clarity, not hype. 𝗪𝗵𝗮𝘁 𝘄𝗲’𝗹𝗹 𝗳𝗼𝗰𝘂𝘀 𝗼𝗻: → Defining where AI brings measurable impact in your organization → Embedding AI into everyday tools (Word, Excel, PowerPoint, Teams) → Applying AI across sales, marketing, finance, operations, support, and engineering → Building simple, no-code automations that actually get adopted → Moving from pilots to repeatable, organization-wide practices 𝗪𝗵𝗼 𝘁𝗵𝗶𝘀 𝗶𝘀 𝗳𝗼𝗿: Leaders who want to start the year knowing how AI will be used, governed, scaled, and measured — not just “explored.” If this resonates and you want to start 2026 with a clear AI direction, you can 𝗿𝗲𝗴𝗶𝘀𝘁𝗲𝗿 𝗵𝗲𝗿𝗲: https://lnkd.in/d28v7Hkn #AI #AIImplementation #AIStrategy #DigitalTransformation #AI2026 #Leadership #EnterpriseAI

7

Today I’m tuning into two online events that both circle the same big question: how exactly is AI reshaping our industries right now? Ciklum DevTalks 2025 SEO Conference “Collaborator 2025” Different domains, same tension in the air — the ground is moving under everyone’s feet. One talk already pulled me in: “Looking Ahead: What’s on the AI Horizon?” It digs into a point many people prefer to gloss over. AI is accelerating, yes. But current LLMs still wrestle with real constraints: data quality, memory, and reasoning depth. Some predict AGI around the corner; others see architectural ceilings we haven’t cracked yet. James Lennon breaks it down with uncommon clarity: • why today’s LLMs hit their limits • how rising token costs quietly influence the future of AI products • why Generative UI might be the next interface shift — not hype, but an economic necessity Following talks from two different industries in one day makes the pattern hard to ignore: AI isn’t just adding speed. It’s rewriting expectations, workflows, and eventually whole markets. More thoughts after the sessions... #AI #ArtificialIntelligence #LLM #AGI #TechTrends #CiklumDevtalks #Collaborator2025 #SEO #FutureOfAI #GenerativeUI

6

Part 3: The Crossroads: When a FlixBus Model Meets a Matatu Culture So, we have two transport models, worlds apart: 1. FlixBus: A centralized, corporate giant thriving in a formal, regulated European market. 2. The Matatu System: A decentralized, informal, and culturally embedded network perfectly adapted to the realities of urban Africa. What happens when you try to impose the first model on the second? You get SWVVL. The failure of SWVVL wasn't a simple case of a startup being fought by incumbents. It was a fundamental clash of systems. It was a story of applying a corporate, top-down solution to a deeply fragmented, bottom-up ecosystem. The problem wasn't the technology; it was the assumption that a different cultural and economic structure was a problem to be "fixed." This brings us to the billion-dollar question: Can a large-scale, efficient, and safe public transport company exist in Sub-Saharan Africa? Yes, but it won't look like FlixBus. And it won't be built by replacing the matatu. The future lies in a hybrid model. A model that respects the entrepreneurial spirit of the matatu owner but provides the structure and technology of a modern corporate entity. It's about partnership, not replacement. Imagine a platform that works with the matatu SACCOs to provide: • Technology: For route optimization, cashless payments, and predictable scheduling. • Capital: For upgrading vehicles and improving safety. • Structure: For better driver training, benefits, and professionalization. This is not about destroying the matatu culture, but about giving it the tools to evolve. It’s about transforming a fragmented system into a powerful, integrated network without losing its soul. This is the real challenge and the greatest opportunity in African transport. The goal isn't to build another FlixBus. It's to create something entirely new, born from the unique realities of the African continent. #FutureOfMobility #AfricanInnovation #Partnership #TransportTech #Matatu #SocialTransformation

42

Model owners like OpenAI are vertically integrating to extract the value delivered to clients in real-life use cases, rather than just trying to top the benchmarks for model capability. It'll be interesting to see how the GTM strategies of the different labs play out over the coming 12 months as we're now very much in the final run to effective AGI in a corporate employee sense. With the big brand labs doing deals such as this (and Anthropic putting in place a formal implementation partner program), will we see adoption being influenced by vendor lock-in / conflicts of interests? It really feels like that some of these labs really want to have their cake and eat it! I do wonder, however, how many organizations that are buying AI transformation services actually understand that they've got the opportunity to flood their businesses with scalable on-demand intelligence which is best applied to wholly reimagined workflows rather than the classic 'make existing employees more productive' mentality.

5

The SOW agreement to delivery should follow a simple logic: ‘Go from A to Z’. But the reality? It usually looks more like: A → B → E → J → F → Z… then back to B. Why? Because when the agreement is not structured, the execution isn't. • Teams have different expectations • Scope lives in inboxes • Milestones are vague • And “done” means different things to different people It’s not that the work is impossible, but if the setup is unclear, minds will scramble. When your business and supply chain are aligned from the start? You don’t just get from A to B, you move smoothly through C, D, E and beyond. Everyone who uses and sees what we have created with OneXRM around simplified SOW creation has confidence that everyone will be on the same page. And when everyone is on the same page… 📍 Less drama, more strategy and action 📍 Less admin, no extra costs, reduced risk of delays 📍 Better delivery & motivated teams It all starts with clarity. Let me and the team show you precisely that - no drama, just 15 minutes of your time, and you'll save much more of your team's time by delivering!

7

🚀 What happens when imagination collides with world-class AI automation—no code required? You get the world’s largest hackathon in history, and it’s happening right now at bolt.new! Over 80,000 participants have already joined this epic, decentralized build-a-thon, all vying for their slice of a $1,000,000+ prize pool and a shot at product stardom. **The Future is No-Code, and It’s Global** The bolt.new hackathon is shattering expectations in the AI and automation space. Forget the cliché of hackathons reserved for elite coders—here, students, creators, full-stack engineers, and idea-first founders are using the bolt.new platform to launch, scale, and demo products that once required months of coding. The barrier to entry has been obliterated: just describe your vision and watch as AI-powered tools build your app across low/no-code platforms. **A Movement—Not Just a Competition** This isn’t your average hackathon; it’s a global launchpad. Community-powered events are springing up worldwide, amplified by partnerships (Anthropic’s Claude Sonnet 4 now powers all Bolt users, unlocking real AI-building potential). The lineup of judges—from indie legends like Pieter Levels to AI investors like Sarah Guo and platform builders like Evan You—signals just how much industry attention is focused on what comes next. As the official event page puts it: “This is your classroom. This is your funding round. This is your moment!” **$1,000,000+ Reasons to Build** Beyond fame and momentum, the prize pool includes over $1M in cash, AWS credits, and cool tech—plus awards for everything from monetary innovation to “Silly Sh!t.” Whether you’re into automation (think RevenueCat, Supabase, Netlify) or want to show off conversational, voice, or blockchain skills, there’s a niche (and likely a prize) just for you. There are even tracks for making your app go viral, or proving unicorn potential, and regional awards across Americas, EMEA, and APAC. **From the Community: Real Builders, Real Results** The participant buzz is palpable, with energy and questions consistently lighting up LinkedIn, Reddit, and Discord. For many, it’s their first build or their first time using AI to ship an MVP. As Product Lead Martin Slaney notes: “Most can’t code and have never built anything before.” But with bolt.new, they’re launching anyway—and so can you. **Are You Ready to Join History?** Tech’s eyes are on this event. If you believe in accessible AI creation and want to ride the next wave of innovation, this is your open invitation. Register at hackathon.dev and see what you can build in days—then share it with a global community that’s rewriting the rules in real time.

5

1 Comment

𝗦𝘁𝗼𝗽 𝗚𝘂𝗲𝘀𝘀𝗶𝗻𝗴, 𝗦𝘁𝗮𝗿𝘁 𝗧𝗲𝘀𝘁𝗶𝗻𝗴 — 𝗔𝗜 𝗕𝗲𝗵𝗮𝘃𝗶𝗼𝘂𝗿 𝗜𝘀𝗻’𝘁 𝗦𝘁𝗮𝘁𝗶𝗰, 𝗮𝗻𝗱 𝗬𝗼𝘂𝗿 𝗧𝗲𝘀𝘁𝘀 𝗠𝘂𝘀𝘁 𝗧𝗿𝗮𝗰𝗸 𝗜𝘁𝘀 𝗠𝗼𝘃𝗲𝗺𝗲𝗻𝘁 AI behaviour is a moving system. If you’re not tracking its movement, you’re not testing — you’re hoping. 💭 Mikko Ruuhamo ’s blog highlights something engineering teams are running into everywhere: 𝑨𝑰 𝒂𝒈𝒆𝒏𝒕𝒔 𝒅𝒐𝒏’𝒕 𝒇𝒂𝒊𝒍 𝒊𝒏 𝒄𝒍𝒆𝒂𝒏, 𝒃𝒊𝒏𝒂𝒓𝒚 𝒘𝒂𝒚𝒔 — 𝒕𝒉𝒆𝒚 𝒔𝒉𝒊𝒇𝒕. A small model update, a retrieval change, a prompt adjustment, and behaviour moves just enough to matter. Testing becomes less about correctness and more about 𝗰𝗵𝗮𝗿𝗮𝗰𝘁𝗲𝗿𝗶𝘀𝗶𝗻𝗴 𝗵𝗼𝘄 𝗯𝗲𝗵𝗮𝘃𝗶𝗼𝘂𝗿 𝗲𝘃𝗼𝗹𝘃𝗲𝘀 𝗼𝘃𝗲𝗿 𝘁𝗶𝗺𝗲. 💫 That’s why 𝐬𝐜𝐨𝐫𝐢𝐧𝐠 𝐛𝐞𝐚𝐭𝐬 𝐦𝐚𝐭𝐜𝐡𝐢𝐧𝐠. Accuracy becomes semantic proximity. Reliability becomes variance across repeated runs. Compliance becomes observable behaviour under constraints. A test suite stops acting like a gate and starts functioning like a behavioural monitor — something that reveals drift, brittleness and unexpected movement rather than certifying a single output. 💭 And then the real engineering question emerges: 𝒘𝒉𝒂𝒕 𝒅𝒐𝒆𝒔 𝒐𝒑𝒆𝒓𝒂𝒕𝒊𝒐𝒏𝒂𝒍 𝒄𝒐𝒏𝒇𝒊𝒅𝒆𝒏𝒄𝒆 𝒍𝒐𝒐𝒌 𝒍𝒊𝒌𝒆 𝒘𝒉𝒆𝒏 𝒕𝒉𝒆 𝒔𝒚𝒔𝒕𝒆𝒎 𝒏𝒆𝒗𝒆𝒓 𝒓𝒆𝒔𝒑𝒐𝒏𝒅𝒔 𝒆𝒙𝒂𝒄𝒕𝒍𝒚 𝒕𝒉𝒆 𝒔𝒂𝒎𝒆 𝒘𝒂𝒚 𝒕𝒘𝒊𝒄𝒆? It seems to come from pairing two strengths. - Automation gives breadth — repeated runs, semantic scoring, drift detection, trend lines instead of snapshots. - Humans bring depth — defining what “good enough” means, spotting the edge cases where failure matters, interpreting borderline outputs where nuance determines risk. 💭 When those layers work together, behaviour becomes observable instead of assumed. Drift becomes visible early. Boundaries become clearer. And trust becomes something built on evidence, not optimism. 💭 AI systems rarely fail loudly; they fail quietly. Modern testing is simply a way of hearing the quiet shifts before customers do.

2

1 Comment

We have some exciting things cooking at OkoRide! In less than a year we’ve gone from 0 to having serving more than 300K rides annually on Bolt platform in Riga. Toda, we have over 100 active drivers across our 30 cars and we’re yet to make our first operational hire! This sort of scaling has been largely enabled thanks to AI (shoutout to Replit and Claude Opus 4.6!). Instead of hiring people for day-to-day ops, we’ve been able to build in-house software and agentic workflows that serve as a backbone for all our ops. But more importantly, we’re building the infrastructure that would allow us to operate thousands of vehicles with maximum uptime. Today, driven by human drivers and hopefully soon autonomous vehicles. To kick things into next gear, we’re about to launch our next city — London! Together with Marten Orgna and Uldis Tēraudkalns we’ll be on the ground this coming week as illutrated by Grok below 😀 If you know anyone we should meet on the ground (ridehailing fleets, potential financing partners, charging operators, EV repairshops, insurance players) — tag them in the comments or feel free to DM! More about how we’re building the fleet operations layer for the autonomous future 👇 https://lnkd.in/dmf2kHNH

102

10 Comments

We sat down with Pekka Pönkänen, QA Competence Lead at Wolt, to talk about one of the biggest challenges in tech: how do you maintain quality when you're scaling fast? 📈🫡 Pekka shares how a quality-first mindset transformed Wolt's mobile release process, why integrating QA early in the development cycle matters, and what skills QA professionals need as AI reshapes our industry. 🎧 Listen now: https://lnkd.in/d7eykVPE #TechEffect #Podcast #QualityAssurance #SoftwareTesting #Wolt

18