Petar Ivanov

When we started writing blogs at nekuda, our chief writing officer :) Idan Levin, pushed to write clearly and authentically. Initially, we leaned heavily on ChatGPT for drafts. But over time, we realized readers value our real thoughts, not masked in AI language. Now, we write from scratch. You’ll probably notice grammar mistakes or occasional typos, but that’s intentional. It signals authenticity. It’s human. Today, I’m sharing a new blog about headless commerce and the future commerce stack we see emerging from a merchant’s perspective: /catalog, /cart, /checkout, and /postorder. These end-points will be key pillars of the next era in commerce, especially with the rise of agentic shopping. We’ve been getting many questions about the future of agentic commerce stacks. Here’s our take, straightforward and transparent. We also discuss the delicate balance between brand representation and AI driven interfaces, highlighting emerging standards like MCP-UI (https://mcpui.dev/). Read our full thoughts here, and let us know what you think: https://lnkd.in/erevTHEn #HeadlessCommerce #AgenticCommerce #MCP #Ecommerce #FutureOfRetail

59

5 Comments

In startups, there's no concept of being on call. At Alma, Everyone is on call. Always. At 2 AM when our document processing pipeline went down, it wasn't the "designated on-call engineer" who fixed it. It was whoever saw the Slack notification first. But here's what's changing fast - AI is reshaping what "on-call" actually means. Traditional on-call: Get paged → SSH into server → Debug logs → Apply hotfix → Write postmortem AI-enhanced on-call: Get intelligent alert → AI suggests root cause → Review proposed fix → Approve deployment → AI writes incident summary A few startups are already building this. The engineer's role is shifting from "human debugger" to "decision maker" btw, if you're an immigrant who's exploring work visa options like O-1, H-1B, EB-1A etc. - get in touch with us at Alma.

80

19 Comments

Before we wrote a single line of code for our product, we launched an open-source project. We knew contracts were broken. But we also knew that another tool to manage the same old documents could only help so much. First, we had to change the documents themselves. So we built a library of standard agreements: simple, modern, fair contracts that didn’t require $10k in legal fees to understand. And we made them free and open source. The idea was simple: if we could help startups close faster with better templates, they’d naturally ask, “can you help us manage these too?” That’s exactly what happened. Our open-source contracts became our best growth channel, a signal of trust, credibility, and actual usefulness before we ever asked anyone to pay us. Sometimes the best way to get paying customers is to give something away for free

30

2 Comments

Still stuck with coding bottlenecks and repetitive dev tasks? Most developers spend countless hours debugging, refactoring, and stitching code together—time that slows innovation and adds friction to real software development. Meet Qoder – an agentic coding platform that doesn’t just autocomplete code, but understands your entire codebase and acts like a true AI engineering teammate. With Qoder, you can co-build features, delegate complex engineering tasks, and deliver production-ready code—faster and smarter. Here’s how it helps you think deeper, build better: ✅ Conversational Pair Programming – Chat with Qoder like a professional engineer and co-build across multiple files, with intelligent completions that anticipate your next move. ✅ Quest Mode (Autonomous Programming) – Delegate multi-step engineering tasks. Qoder analyzes, plans, codes, and tests—while keeping you in control. ✅ Repo Wiki & Architecture Insights – Uncover hidden dependencies and patterns in your codebase, turning tribal knowledge into actionable intelligence. ✅ Self-Evolving Memory – Qoder learns your preferences, past solutions, and coding style—becoming more effective with every interaction. ✅ Developer Toolbox – Equipped with built-in tools and MCPs, Qoder can observe, plan, and execute tasks like a real engineer. No more juggling between tools or chasing bugs. Just delegate, collaborate, and ship with confidence. 👉 Use Claude4 for free: https://lnkd.in/dKHkX5nx #QoderAI #ThinkDeeperBuildBetter #AgenticCoding #RealSoftware #AIEngineer #CodeWithQoder

150

21 Comments

Looking at this market map, you might wonder: with all these tools available, why did we build Maestro? The answer lies in what most "agents" actually do versus what enterprises actually need. Too often, what's being sold as "agents" are glorified retrieval wrappers—RAG chatbots that answer questions by pulling up documentation snippets. They're useful for knowledge lookup, but they're not doing knowledge work. The real opportunity is automating high-stakes, multi-step tasks like M&A due diligence, regulatory compliance gap assessments, or contract risk analysis across multiple vendor agreements—not just answering a question. To get agents that actually do knowledge work, most enterprises face a dilemma: Spend months on integrations to build & scale their own stack or buy out of the box solutions. It’s losing time or losing the ability to make the agent work for the specific enterprise environment, data and tasks. This is why we built Maestro, AI21's knowledge agent platform. Maestro breaks down complex knowledge tasks into executable steps on the fly, then dynamically utilizes the best models and tools for each stage, with built-in verification and validation baked into every step. Maestro handles the planning, orchestration, and observability while Jamba models provide long-context capabilities and you can bring whatever other models you prefer. Everything can be privately deployable for secure, efficient, and fully customized knowledge agents. We're proud to build tech that spans every level of the knowledge agent stack, making it easier for developers to create and scale agents for real enterprise knowledge work. Here’s to less chaining, more building. 💪🚀🧩

66

3 Comments