Nillow:// AI research & development
AI research leaves a trace.
Public notes from Nillow's AI R&D team: the concepts, experiments, evaluations, and prototypes behind new AI systems, made inspectable without flattening them into slogans.
- 008R&D note · v1.0Software Is Not Dead. We Have Barely Invented It.AI coding agents will do more than write code. They will push software above implementation—toward world models, ontology engineering, synthetic languages, and executable behaviour.Ontology Engineering / AI-Generated Software / Software Engineering / AI Agents / Synthetic Language Engineering / Cognition Engineering→
- 007R&D note · v1.0AI Won’t Just Replace Jobs. It Will Create a New Profession.Meet the cognition engineer: the profession forming above AI.Cognition Engineering / Artificial Intelligence / Future of Work / AI Systems / Human-AI Systems→
- 006R&D note · v1.0The AI Agent Is the Wrong Unit of IntelligenceWe copied the org chart into the prompt and called it intelligence. The machine deserves a better architecture.Agentic AI / AI agents / AI architecture / Multi-agent systems / Synthetic cognition / Cognitive topology / Agent sprawl→
- 005R&D note · v1.0AI Is Liquefying Information. Consulting Was Built on Information Asymmetry.The next consulting economy will not reward firms that merely know more. It will reward firms that convert liquid knowledge into executable futures.AI consulting / future of consulting / information asymmetry / information liquefaction / execution architecture / cognition engineering / enterprise AI / agentic AI / workflow redesign / AI governance / knowledge work / future overhang→
- 004R&D note · v1.0The Next Cybersecurity Boundary Is InterpretationAI agents can turn stolen fragments into reconstruction plans. Protected systems need semantic warrants, not just file permissions.AI security / Agentic AI / AI-assisted semantic reconstruction / Semantic Warrant Layer / Prompt injection / Model Context Protocol / AI provenance / AI supply chain security→
- 003R&D note · v1.0AI Agents Need a Runtime, Not Another PromptHallucination, vibe coding, MCP, and AI workflows all point to the same missing layer.AI agents / Agentic AI / AI runtime / AI hallucination / Vibe coding / Model Context Protocol / AI workflow automation / AI governance→
- 002R&D note · v1.0What Biology Can Teach Us About Artificial IntelligenceBoundaries, feedback, and layered organization before the brainBiological intelligence / Artificial intelligence / AI architecture / Evolution / Multicellularity / Distributed intelligence→
- 001R&D note · v1.1Every AI Benchmark Has an ObserverHow to read model scores without mistaking the test for the intelligenceAI evaluation / benchmarks / measurement / model selection / LLM evaluation→
AI R&D ENGAGEMENT
Need an AI research and development team?
Nillow researches, prototypes, and evaluates new AI systems. Bring us the capability you need, the uncertainty blocking it, and the environment where it must work.
