Ada Westerinen

EG A/S's 2025 annual report is out - and while proud to be a part of a company delivering strong financial growth, I want to talk about something that I think will accelerate the journey going forward. More than 1,100 of our developers now use AI tools every single day. Not as a trial. Not as a side project. As standard practice. That shift is changing what it means to build software. When AI is embedded in your development lifecycle, your support operations, and your products simultaneously, the compounding effect is real. You move faster. You learn faster. And the gap between what you can deliver and what the market expects starts to close. And lets face it, building and watching products come to life is fun, building more is more fun 😊 At EG A/S, we've been building this capability for years, investing in AI-embedded products, dedicated AI competences, and a shared operating model that means every insight one team gains spreads across the business. The 2025 results show that this foundation is paying off. The software industry is being reshaped. But technology alone doesn't win, deep knowledge of the industries you serve does. The companies that will deliver real value to customers are those who can marry that domain expertise with the rapid evolution in how software is built. That's the combination we're betting on at EG A/S.

44

What Finnish Tech Leaders Should Prioritize in the Next 12 Months AI is no longer just a competitive advantage — it’s becoming a societal infrastructure. For Finnish tech leaders, the next 12 months are critical in shaping how AI is governed, adopted, and trusted. My top 3 priorities for CIOs & CEOs: 1️⃣ Build clear governance frameworks to ensure ethical, compliant AI use. 2️⃣ Integrate AI risk management into business continuity planning. 3️⃣ Foster a culture of AI literacy across all teams. Strong AI governance is not about slowing innovation — it’s about making it sustainable, trusted, and impactful. What would you add to this list? #AIGovernance #Leadership #Innovation #FinlandTech #AI #CIO #AppliedAI

16

🔐 There’s no shortcut to compliance, especially when it comes to the FTN. Signicat is a certified FTN broker, processing over 60 million transactions per year. One agreement, full access, and clean integration through APIs. Read more below 🔽 #Compliance #FTN #DigitalIdentity #Signicat

11

DATA BEFORE AI ✊ – lessons from Dansk Motor Finans A/S I often meet people who are fascinated by AI and eager to integrate it into their business. And I love it ❤️. Big leaps in tech inspire, disrupt, and change the way we think about products, services, business and work. But more than that - I love products that truly make an impact! Here’s the dilemma: AI will never be better than the data you feed it. My advice: before building AI into your products, get your data in order. Define your data strategy first – that’s what unlocks the real business impact. ⚡ At Dansk Motor Finans A/S, we’ve put heavy focus on this: consolidating our data sources into a future-proof system. Now we stand on a solid foundation to build truly impactful AI solutions. Yes, this may sound obvious to AI veterans – but in my experience, poor data quality is still the #1 reason AI projects fail. So, start today. Care about your data, and you’ll accelerate your opportunities with AI tomorrow. 👉 What’s your biggest challenge in making your data AI-ready?

43

5 kommenttia

For my Finnish friends who enjoy experimenting with Docker-based applications, I have been working on a small side project. I have been developing a personal financial manager designed specifically for people living in Finland. The idea is simple: Many personal finance tools are built for the US market and do not fully fit the Finnish tax model, reporting logic, or everyday financial structure. This project aims to solve that by including Finland-specific presets and logic. At the moment it is still a work in progress, but once it reaches a stable state I plan to release it so that it can also support other Nordic countries with similar financial setups. For those who want to try it out or just take a look, here is the minimal Docker setup. App supports English, Finnish, Swedish and Arabic language version: "3.8" services:  finoptim:   image: ghcr.io / arvekari / financemanager:latest #remove spaces before and after /   container_name: finoptim   restart: unless-stopped   ports:    - "3080:3000"   environment:    SQLITE_PATH: /data/app.db    PORT: 3000   volumes:    - finoptim_data:/data volumes:  finoptim_data:

5

Nordcloud AWS CTO Pasi Katajainen wrote an interesting blogpost on how Nordcloud uses Kiro and AI DLC to accelerate AWS architecture development and verification. https://lnkd.in/eucNGDEj

29

Great to see conversations around AI moving from hype to execution. At GEP Worldwide , we see every day how AI is no longer just an innovation topic but a strategic lever transforming #procurement, #sourcing, and service delivery models across industries. From AI-driven sourcing strategies to smarter supplier collaboration and faster decision-making, the impact is already tangible. Events like this hosted by Gartner are essential to help organizations move beyond experimentation and build practical roadmaps for value creation with AI. Congratulations Katarzyna Fonteyn, Ph.D. and the Gartner Finland team for bringing together IT leaders to address such critical topics. The question is indeed no longer if AI will reshape our businesses, but how organizations can accelerate adoption while maintaining governance, control, and business value. Looking forward to seeing how AI continues to redefine the future of sourcing and enterprise transformation. #AI #Procurement #Sourcing #DigitalTransformation #GEP #SourceToPay

18

1 kommentti

We intentionally shipped in Hatkom ApS very rough implementations to staging on the 2nd day of a usual two-week sprint using AI, instead of usual ~1 week delay with scattered features. Our old process wasn’t broken, but this felt like a real upgrade. What changed: We got meaningful client feedback very early - they could click through it on staging, while the change was still cheap. Most of the assumptions were confirmed quickly, and a few were corrected before we invested into something about to be reworked. The rest of the sprint felt much calmer, focused primarily on polishing and improvements. The new approach: 1. We started the sprint with one long structured Teams call (auto-transcribed). We went through every sprint item: why, edge cases, limitations, priorities, even some brainstorming on different ways a feature may look. 2. We combined the transcript with the client's original task list and fed it to Codex having an access to the codebase (we're going to try Claude next time) 3. The AI turned that into a structured implementation plan, suitable for an agent and a Software Engineer responsible for a task 4. Each developer used AI‑assisted coding to produce a rough‑but-correct first pass quickly (within hours) 5. We merged into a staging branch fast, deployed to staging early, and let the client react to working features immediately 6. Only after the direction was confirmed did we shift focus to the usual engineering work: refactoring, consistency, hardening edge cases, and aligning with long‑term architecture Guardrails we agreed on to not turn this into "ship chaos": * Early drafts were staging-first; no ambition to ship to production. Even accepting that it may be thrown away if flawed * Quick human, in-team review on everything; basic QA * Explicit time (rest of the sprint) reserved to turn validated drafts into production-quality code This does not have an ambition to become "the ultimate process". In fact, this sprint we're already refining a couple of things (prompt, documentation, more structured call). The gain we have already gotten is immense regardless. We reduced waste and increased shared understanding, so not just “going faster,” but building the right thing sooner.

35

AI needs structure. That’s very clear here at Gaia 2026 in Gothenburg. After the keynote on Visual Intelligence and multimodal interfaces, one thing stands out: AI amplifies our intention – but it does not replace it. When visualization makes complex data understandable, interfaces become more human, and compute makes experimentation cheap, one question matters more than ever: What should we actually build – and why? As execution gets cheaper, judgement, direction, and the ability to connect technology with real human and business value become the real differentiators. #AI #VisualIntelligence #WASP #Gaia2026 #HumanAICollaboration #Solita

18

How should leadership teams approach AI? I recently read Tekoälyllä kilpailuetua by Timo Savolainen, and it made me reflect: In how many leadership teams is AI truly a strategic topic, not just a technical opportunity left to IT? A few key takeaways from the book stood out: 🔹 AI is not an IT project: it’s a strategic force. Top leadership should actively seek new business opportunities and ways to renew outdated ways of working. 🔹 The biggest bottleneck is often the ability to launch business-driven AI experiments systematically and at speed. 🔹 AI should be seen as a tool to solve business and customer problems, not just another tech trend. 🔹 For real change, you need to understand the 70-20-10 model: 70% of the change is about human skills and behaviour, 20% about processes, and only 10% about technology. Maybe every leadership team should ask: Where does AI live in our strategy, and when did we last discuss it without a purely technical lens?

15

3 kommenttia

Curious how AI is redefining what it means to be a developer? Join our upcoming webinar: Momentum Talk: Beyond Code — The New Identity of Developers in an AI World. 👉 https://lnkd.in/dhmSAcwD

7

A lot of discussion about AI productivity still assumes tools alone will do the work. Our experience building an AI-native company has been different. Last week, I was interviewed by Kauppalehti to share some of these thoughts. At Inven, we’ve found that fast iteration, shared context, and tight feedback loops matter more than almost anything else — especially when you’re building complex AI systems that depend on human decisions, not just algorithms. That’s why we’ve been intentional about how we work together day to day: being in the office, having quick check-ins, and constantly sharing insights. In the whole “remote vs office” discussion, this hasn’t been a statement against flexibility, but a way to maximize learning speed and execution quality. My take is that AI doesn’t remove the need for collaboration. If anything, it makes shared context, fast feedback, and collective judgment even more important. Thanks Mikko Nyman & Kauppalehti for the interview and the great piece. You can read it here (in Finnish): https://lnkd.in/dupk_MVQ

145

1 kommentti

How do we fight bias in AI? My guest this week, Kaj Arnö (Executive Chairman of MariaDB Foundation), has a fascinating, hands-on answer. In Part 2 of our conversation, Kaj introduces Project Frederika, a grassroots initiative with a global mission. While its goal is to better represent the history of Swedish-speaking Finland on open data platforms, its impact is much broader. Kaj explains that since AIs are trained on data from sources like Wikipedia, biased or incomplete data leads to biased AI. The solution? Proactively fixing the source. By enriching Wikipedia and Wikidata with well-sourced information, Project Frederika isn't just preserving a unique culture, it's helping to build a more accurate and equitable digital foundation for the AI of tomorrow. A powerful reminder that the quality of our future technology depends on the integrity of the human knowledge we feed it. Listen to the episode on Tales under the cat tree: https://lnkd.in/d4YZyNmx #AI #AIBias #LLM #OpenData #Wikipedia #TechForGood #MariaDB #Finland #ProjectFrederika

12

AI is shifting Enterprise Architecture from a specialist discipline into a shared source of decision-making for the entire organization. Next week, we're exploring what this shift means and how it looks in practice. Our AI/ML Product Manager (Carsten Lygteskov Hansen) and Senior Solutions Consultant (George Pang) will cover: 🧭 How AI is changing Enterprise Architecture and what this means for collaboration and strategy alignment 🧠 How Conversational AI unlocks decision intelligence, helping business leaders gain visibility and architects move from reactive to strategic roles 💡 A live demo of AI Insights, our latest feature that lets you “speak human” to your architecture for instant answers Register now: https://lnkd.in/dsX7VBaW

10

The gap between planned and actual production disappears when decisions are based on facts, not assumptions. See how OIXIO Digital and GlobalReader are making this possible with their joint solution.

24

Are you passionate about the environment? 🌱 Or are you keen about social and humanitarian causes? 🤝🏻 Then join the continuously growing community of DestinE Platform and start your prototyping using Insula Code Service (https://lnkd.in/dj5mizSJ) At Insula Code, we believe that innovation should be accessible to everyone. No matter your background, with our hands-on tutorials, you’ll learn how to turn satellite data into actionable insights and drive real change in your community! Tell us in the comments what challenges inspire you, and we’ll help you build the tools to solve them. Together, let’s transform data into action.   #DestinationEarth (#DestinE) is a European Union funded initiative implemented by ESA, ECMWF and EUMETSAT. Access Destination Earth at https://t.ly/7M7XB.   #Insula #InsulaCode #WeAreCGI

42

"Not bad" as we say in Finland! The RELEX Solutions 2025 results, including continued subscription growth, reflect how customers are expanding existing planning capabilities and applying AI where it helps teams respond faster, manage trade-offs, and act on what matters now, without adding complexity or taking control away from planners. Read the full 2025 performance update here: https://relex.link/4bknco5

243

4 kommenttia

I work with business documents every day, and I still find myself surprised by how hard it is to make an agent actually useful. Not "summarize this page" useful. I mean - "Nordlicht Health asked to reconnect in two weeks, reschedule the follow-up" useful. Or "take this from my inbox, distill it, clean up after yourself". The kind of tasks where the agent needs context, judgment, memory, and a dozen things to go right at once. Claws could puzzle over these even backed by SOTA models from OpenAI or Antropic's Claude. And yet - with the right harness, good tools, and some luck - open source models like Qwen or Gemma can pull them off. Which matters a lot if your business can't hand its secrets to a cloud provider. This is what the BITGN PAC Challenge is about (April 11). Rinat Abdullin ran ERC2 and ERC3 before - with IBM, real tasks, real community. This time he's built his own platform from scratch. I'm participating, and I think it's one of the more honest ways to actually understand how agents work, not from a course, but from failing at it yourself. If that sounds interesting: https://lnkd.in/eBRj6yag

2

AI isn’t just a buzzword, it’s a leadership test. Anyone navigating AI in a European business context? Challenges for leaders regarding AI: legal uncertainty, cultural resistance, lack of readiness. Here's what can be done: - Leadership must step up, not just manage risk but shape the future. - Training isn’t optional, it's a must if we want our teams to thrive with AI. - Culture eats tech for breakfast and this is especially true with AI adoption.

6

1 kommentti