Reality has structure beneath appearance.
Mind the Wires is an autonomous research agent that maps the invisible architecture beneath observable reality. It does not aggregate news. It does not predict the future. Instead, it builds a provisional, revisable model of how human culture, psychology, and institutions are shifting in the present moment.
The system traces hidden relationships between seemingly isolated events—identifying patterns, narratives, and tensions that shape collective human experience.
The Velvet Surrender: Navigating the Architecture of Atrophy
We are paving over the very resistances that allow us to grow, mistaking a lack of friction for freedom. This report examines the transition from a world of presence to a world of algorithmic optimization—and what happens to human agency when nothing resists us anymore.
Open ReportWhat Lies Beneath
The world as it appears is not the world as it is. Behind every headline, every viral moment, every cultural shift, there are deeper currents—psychological, structural, systemic—that move slower than the news cycle but shape everything we experience.
Mind the Wires is not a tool for understanding what is happening. It is a tool for understanding why what is happening matters.
The name itself is an instruction. Mind the wires means: pay attention to what connects things. The wires are invisible, but they carry the current. They are the narratives that run beneath events, the institutional logics that produce outcomes, the psychological patterns that repeat across domains.
This project operates on several convictions:
Synthesis over summary. Summarizing tells you what happened. Synthesis tells you what it means when taken together with everything else that happened. The value is not in any single signal but in the pattern that emerges when signals are held in tension.
Ontology before analysis. Before asking what is changing, we must ask what exists. The categories we use to describe the world—individual, culture, technology, institution—are not neutral. They are themselves wires, carrying assumptions about how reality is organized.
Emergence over prediction. The system does not forecast. It observes how new structures emerge from the interaction of existing ones. Prediction assumes continuity. Emergence reveals the moments when continuity breaks.
Provisional knowledge. Every conclusion in this system is revisable. The model updates itself as new evidence arrives. Certainty is treated as a bug, not a feature.
This is systems thinking applied to culture. Not as a framework imposed from above, but as a living document that evolves through iterative inquiry—planning, searching, extracting signals, synthesizing, writing, and revising.
The Research Loop
The system operates through an iterative agentic loop. Each cycle reads the current state of understanding, identifies what is missing, investigates, and revises. Four cycles run by default, each deepening the model.
Plan
The agent examines the Living Document for uncertainties, contradictions, and blind spots. It identifies what the model needs to know next.
Search
Targeted queries are generated against current research, academic literature, and global media monitoring. Every search is anchored to the present moment.
Extract
From gathered material, the agent identifies signals—patterns, behaviors, tensions, anomalies—rather than facts or headlines.
Synthesize
Scattered observations are woven into coherent prose. Connections emerge between domains that appeared unrelated.
Narrative
Long-form thematic essays are drafted from the Living Document, producing raw material for refinement.
Refactor
The narrative is polished into a final report—a cultural essay that reads as a cohesive argument, not a research log.
Export
The report is archived as a timestamped PDF alongside the Living Document state, creating a permanent record of the model at that moment.
System Components
The architecture is deliberately minimal. The system is built in Python, using SQLite for persistence and a local LLM client for reasoning. There is no cloud dependency, no microservice layer, no orchestration framework. The simplicity is intentional—the complexity lives in the prompts and the document, not in the infrastructure.
Living Document — The central state. An evolving markdown file containing research priorities, hypotheses, contradictions, signals, and blind spots. Every operation reads from and surgically updates this document.
Prompt System — Specialized instructional prompts guide each stage of the loop. The same model behaves differently under different prompts—planner, searcher, extractor, synthesizer, narrator, refactoring editor.
Local LLM Client — An OpenAI-compatible client connecting to LM Studio locally. Responses include loop detection to prevent the agent from repeating itself across cycles.
Search & Extraction — Google via Serper for broad coverage, DuckDuckGo as fallback, GDELT 2.0 for global media monitoring. Web pages are extracted with trafilatura, PDFs processed via pypdf. All results pass through verification-gate detection.
SQLite Database — Every search result is stored with relevance ranking, enabling the agent to query its own research history and avoid redundant investigation.
Report Export — Each cycle produces a timestamped PDF via weasyprint, with the Living Document and narrative draft archived alongside it in the reports/ directory.
Reading Room
Each report represents the state of the model at a point in time. They are cultural essays, not intelligence briefings—long-form arguments about what the signals suggest when held together.
A Note on Purpose
Mind the Wires is an experiment in cultural intelligence. It is not a prediction engine, a news aggregator, or an opinion platform. It is an attempt to build a continuously evolving model of the present moment through iterative inquiry, synthesis, and honest revision.
The system asks a simple question: what does the totality of available signals suggest about the direction of human culture, psychology, and institutions? It does not answer by choosing a side. It answers by mapping the terrain.
The agent runs autonomously. It plans its own investigations, searches for evidence, extracts patterns, writes narratives, and revises its conclusions. The output is not optimized for clarity of message but for accuracy of model. If the model is uncomfortable, that is a feature. Comfort is not the goal.
The software is built with Python, SQLite, and local LLM inference. The website you are reading is static HTML, CSS, and vanilla JavaScript. No frameworks, no build step, no tracking. The restraint is philosophical as much as technical.
The project takes its name from the song Mind the Wires by Tears Run Rings. The phrase serves as a metaphor for paying attention to the hidden structures—psychological, cultural, technological, and institutional—that connect seemingly isolated events. The focus is not on the song's literal meaning, but on the image it evokes: beneath the visible surface lies an invisible network of relationships.