Building autonomous AI systems — where natural language is the only interface.
Testing Virtual company Netanel Systems — an AI native company where every function runs on agents. Engineering, content, QA, delivery — all autonomous. No human in the loop unless something breaks.
Right now I'm deep in:
- Autonomous agent systems — multi-agent pipelines that run in production, unsupervised, daily
- 100% AI-native development — no manual coding. Everything written through natural language and agent collaboration
- Model philosophy & behavior — testing how different models reason, where they fail, how to build reliable systems on top of unreliable components
- Agentic infrastructure — building the company OS: memory graphs, orchestration, self-improving agents, coordination protocols
This is not prototype work. These systems run every day and get smarter with each cycle.
Python LangChain LangGraph Claude GPT-4o FastAPI Graphiti FalkorDB Playwright Docker
vision-assistant — Edge AI agent for home use. Local face recognition, YOLO detection, GPT-4o Vision. 25-30 FPS on consumer hardware. No cloud, no telemetry.
I believe the next generation of software is written by AI, orchestrated by AI, and improved by AI — with a human setting direction.
I'm testing that belief in production, every day.
- Dev.to: @klement_gunndu
- LinkedIn: Klement Gunndu
Natural language. Autonomous systems. Real production.