I've been trying to figure out how to show someone what ICM actually looks like in under 30 seconds. You know how it goes — you start explaining numbered folders and stage contracts and human review surfaces, and their eyes glaze over around "stage contracts." So I built a demo that fits on a business card. Tap it to your phone, type what you do for work, and a live three-stage ICM pipeline runs in your browser. You watch each stage complete, read the output, pick what you want built, and walk away with a real workspace folder tailored to your work. All in about 20 seconds. ## The NFC Part I put an NFC chip on the card. That's it. Each chip stores a short random ID — think license plate number. Tap the card, your phone reads the ID, sends it to a server, and the demo starts. I went with NFC over QR because tapping is just easier than framing with a camera. Also lets me tag each chip with context — I can attach "SF Tech Conference — dinner" on the backend and it shows up as a badge when someone taps. The URL carries nothing personal, just the random ID. The chip isn't the point though. It's just how someone triggers the demo. ## How the Pipeline Works Someone taps the card. Their browser opens to a text box: "What do you do for work?" They type something. Then three stages run, each feeding into the next: **Stage 1 — Discovery.** AI looks at what they typed and writes a quick analysis of their work and what parts are repetitive. **Stage 2 — Mapping.** AI designs a multi-step workflow for them. Each step has a note about what they'd want to review before moving forward. **Stage 3 — Opportunities.** AI suggests four specific projects they could build with ICM. Each stage streams back in real time. You can read what the AI produced before the next stage starts. That's the ICM thing — every output is visible and reviewable. Then they pick a project and the AI generates a real workspace. ## What They Walk Away With A folder of markdown files. Usually 5-7 of them. Numbered stages, instruction files per stage, a getting started guide. Download it, open it in Claude Code, type `setup`, and go.

I started by searching our community and didn't find anything. As I learned today, this GitHub repo has blown up 0 - 60K stars in less than 6 weeks. UNDERSTAND ANY CODEBASE IN MINUTES! https://github.com/Egonex-AI/Understand-Anything Graphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more. --------------------------------------------------------------------------------------------------------- EXPLORE THE STRUCTURAL GRAPH Navigate your codebase as an interactive knowledge graph — every file, function, and class is a node you can click, search, and explore. Select any node to see plain-English summaries, relationships, and guided tours. UNDERSTAND BUSINESS LOGIC Switch to the domain view and see how your code maps to real business processes — domains, flows, and steps laid out as a horizontal graph. ANALYZE KNOWLEDGE BASES Point /understand-knowledge at a Karpathy-pattern LLM wiki and get a force-directed knowledge graph with community clustering. The deterministic parser extracts wikilinks and categories from index.md, then LLM agents discover implicit relationships, extract entities, and surface claims — turning your wiki into a navigable graph of interconnected ideas. FEATURES 🧭 Guided Tours Auto-generated walkthroughs of the architecture, ordered by dependency. Learn the codebase in the right order. 📊 Diff Impact Analysis See which parts of the system your changes affect before you commit. Understand ripple effects across the codebase. 🏗️ Layer Visualization Automatic grouping by architectural layer — API, Service, Data, UI, Utility — with color-coded legend. 🔍 Fuzzy & Semantic Search Find anything by name or by meaning. Search "which parts handle auth?" and get relevant results across the graph.