Gregor Ojstersek

One of the talking points of our latest 𝐆𝐫𝐞𝐞𝐧 𝐈𝐓 session led by our colleague, Mihnea Andrei Petrescu, backend engineer, was 𝑒𝑣𝑒𝑟𝑦𝑑𝑎𝑦 𝑎𝑐𝑡𝑖𝑜𝑛𝑠 𝑓𝑜𝑟 𝑎 𝑚𝑜𝑟𝑒 𝑠𝑢𝑠𝑡𝑎𝑖𝑛𝑎𝑏𝑙𝑒 𝑑𝑖𝑔𝑖𝑡𝑎𝑙 𝑤𝑜𝑟𝑙𝑑. We've compiled and explained some simple and effective habits worth sharing: 🤳🏼𝐂𝐡𝐨𝐨𝐬𝐞 𝐚𝐮𝐝𝐢𝐨 𝐨𝐯𝐞𝐫 𝐯𝐢𝐝𝐞𝐨 when it comes to calls. Video requires a lot more bandwidth and energy. When video is a must, 🛑 avoid blurred & artificial backgrounds - rendering increases your device's processing load. 🔗 𝐔𝐬𝐞 𝐥𝐢𝐧𝐤𝐬 𝐢𝐧𝐬𝐭𝐞𝐚𝐝 𝐨𝐟 𝐚𝐭𝐭𝐚𝐜𝐡𝐦𝐞𝐧𝐭𝐬. Uploading files to SharePoint and sending a link saves storage space and avoids duplicating heavy attachments across multiple inboxes. 🌌 Try dark mode. Dark pixels typically use less power on OLED screens. 💤𝐏𝐮𝐭 𝐲𝐨𝐮𝐫 𝐝𝐞𝐯𝐢𝐜𝐞 𝐭𝐨 𝐬𝐥𝐞𝐞𝐩 after 15 minutes of inactivity to avoid energy waste. 🚮 𝐒𝐜𝐡𝐞𝐝𝐮𝐥𝐞 𝐫𝐞𝐠𝐮𝐥𝐚𝐫 𝐝𝐢𝐠𝐢𝐭𝐚𝐥 𝐜𝐥𝐞𝐚𝐧𝐮𝐩𝐬. Removing unused files from your device and SharePoint reduces storage demand and helps data centres use less energy.

17

1 Comment

We recently analyzed Google Trends data for Romania, looking specifically at AI searches over the past year. The results defy the usual assumption: people aren’t turning to AI for productivity or efficiency. Instead, the fastest-growing queries are for tools that sing, draw, or generate video, names like DeepSeek, Suno, and Sora. AI here is treated less like enterprise software and more like a creative sandbox. The data shows two clear patterns. First, curiosity is concentrated on generative and entertainment-driven tools: text-to-image models, music generators, video platforms. Second, there’s a sharp rise in searches for “AI content detection.” This reveals an underlying perception problem: many still see AI as a shortcut, even a form of cheating. That stigma becomes a barrier when thinking about AI in the workplace, if employees and managers equate AI with dishonesty, they will hesitate to adopt it for legitimate tasks. That curiosity in the creative space is healthy, but it hides a risk. If your only encounter with AI is a music generator, you might conclude the technology has no role in serious work. It’s like judging electricity by a novelty lamp, or cars by a go-kart ride, amusing, but misleading. For businesses, consumer AI is the spark, but not the standard. True transformation comes when AI is designed around your own processes and data. Experimentation creates curiosity. Integration creates value. #AIInnovation #CreativeTech #GenerativeAI #DigitalTransformation

13

We’re about to kick off something really exciting, and I wanted to share it with you. MHT EXPERIENCE is officially the first company in our 𝐍𝐞𝐨 𝐱𝐀𝐈 pilot program. We’re excited to take this step in turning AI from “buzzword” to business reality for Romanian companies. MHT works in the electrical infrastructure sector and they’re looking to improve one of the most time-consuming and costly parts of their business: creating commercial offers. Right now, it takes them up to 4 weeks to prepare a proposal. We’re going to help them bring that down to just 5–7 days. 𝐖𝐡𝐲 𝐝𝐨𝐞𝐬 𝐭𝐡𝐢𝐬 𝐦𝐚𝐭𝐭𝐞𝐫? Because time is money. And in a sector with tight margins and fierce competition, speed = opportunity. 💡 According to a McKinsey survey cited in Stanford’s 2024 AI Index, 59% of companies already report increased revenues from AI use. But many still get stuck in “pilot purgatory.” For MHT, we’re building a custom AI solution that will: •⁠ Pull in client and supplier documents (PDFs, Excels, emails, you name it) •⁠ Extract the key data •⁠ Apply the necessary pricing rules and commercial logic •⁠  Generate proposals that are ready for review The goal is simple: 𝐭𝐨 𝐡𝐞𝐥𝐩 𝐌𝐇𝐓 𝐦𝐨𝐯𝐞 𝐟𝐚𝐬𝐭𝐞𝐫, 𝐰𝐢𝐭𝐡𝐨𝐮𝐭 𝐬𝐚𝐜𝐫𝐢𝐟𝐢𝐜𝐢𝐧𝐠 𝐪𝐮𝐚𝐥𝐢𝐭𝐲. AI will speed up the process, but the final decision-making stays with the team. The great news for you? We’re just getting started. MHT is just the first of 10 companies we’re working with in this program. If you’re a Romanian company with €3M–€50M in revenue, and you’ve got time-consuming internal processes you want to optimize, 𝐭𝐡𝐢𝐬 𝐜𝐨𝐮𝐥𝐝 𝐛𝐞 𝐟𝐨𝐫 𝐲𝐨𝐮. Link to the Forbes article in the comments 👇

23

1 Comment

Serbia: An outsourcing hub… but can it become a product nation? . . . . . For years now, Serbia has been known as a reliable outsourcing destination: strong engineers, cost efficiency, solid delivery. But here’s the challenge > when agencies or project-based companies decide to build their own product, many hit the same wall: - No product mindset (thinking in features, not in value). - No go-to-market strategy (amazing tech, no paying users). - No iteration culture (projects are delivered once, products are never “done”). Outsourcing builds skills. Product building requires vision, ownership, and customer obsession. If Serbia wants to move from “workbench of Europe” to “product powerhouse,” the shift in mindset is inevitable. Do you see Serbian companies making this transition?

47

13 Comments

If you pivoted from custom software development to build your own product because of FOMO, congrats, any solid agency can now clone your sass with mid diff. I was right. Last year, I argued, despite the hype. that software companies should double down on custom software services while everyone else insisted they should build their own products for the sake of “innovation in Romania.” Then, long story short, gen AI accelerated, and vibe-coded products rolled out to million users before any of those hundred headcount software companies shipped anything. My reasoning was simple and anchored in pragmatism: - Delivery CAN be engineered: it’s part systems, part training and experience Innovation, on the other hand, happens in garages, not in corps. You can’t guarantee lab-grown vision - Innovation would eventually emerge naturally and when it does, it will need to meet a backround of foundational operational excellence. But this is prime time for both founders and software agencies: - AI has somewhat matured: we now know what’s achievable, what’s not, how much it costs, how fast it runs, and whether users actually want it. - Frameworks have stabilized: just like Rails did for Ruby or Spring for Java, AI now has its own ecosystem (LangChain, LlamaIndex, etc.). We don’t have to reinvent every integration. - Infrastructure is abstracted: between Kubernetes templates and cloud services such as AWS’s Sagemaker, deployment isn’t the bottleneck anymore. Time to production is back to hours -Clients are pragmatic: they no longer say “let’s just do something with AI” or ask for a Copilot. Instead, they make more specific requests: semantic search, knowledge graphs, compliance monitoring, etc - Capital is cautious again: investors want traction and clear ROI again, vague promises of innovation won’t cut it anymore, giving power back to builders who can actually deliver. So great artisans still win. If you need a team that ships with care and clarity, reach out. https://lnkd.in/d3QkiZuy

12

1 Comment