Blog

🤖3 years ago I’ve been starting this blog and made a promise avoding AI content here. Things are changing, yet I don’t want to give up on the promises. I’m introducing “ai-augmented” tag that will be attached to all upcoming posts that was processed heavily by AI. You don’t have to guess - decide what matters to you. My Hermes agent hit the memory cap. So I built a second layer. Most agents stop at one. Mine has two. Using the Google’s OKF specification. The first layer is a 2,200-character f

AI agents are the new drill in the software engineer’s toolbox. Faster? Absolutely. Revolutionary? Maybe. But a drill doesn’t replace the process of building a house. In this article, I explore why SDLC, architecture, testing, and delivery remain as important as ever.

I run two agentic AI setups in parallel - OpenClaw and Hermes. Both on disposable VMs on a Proxmox host in my office. Here's what the stack actually looks like, what works, and where it still breaks.

I thought the hardest part would be latency. We were building a realtime conversation assistant: a system that listens to speech, understands conversational context, and generates suggestions fast enough for someone to actually use them during a live conversation. On paper, the pipeline sounded straightforward:speech-to-text → context processing → LLM generation → suggestions. In reality, everything broke once humans started talking. It went even worse when tested in the middle of crowded ope

An event that happened this weekend was really great. Overall it was chill, friendly and felt like the main goal of the hosts was to provide a platform. Rules? Yes, bare minimum. Guidelines and flexibility were dominating. And it reduced the atmosphere of competition. And guys were gathered around their project and ideas. Building. And the finale reminded me hard truth: I can do anything, but not everything. In 24 hours we have built a local-first working real-time speech assistant that listen