What surfaces on the second read of Anthropic's Advisor Tool. This isn't new architecture — it's a temporary fix shaped by 2026 pricing. A pattern that disappears once Opus prices drop, and eleven other papers from the same period are quietly moving the same way. The anchor of the series.
How do you organize the sliced agents. Four structures, paired with one skeptical paper that argues 'LLM swarms aren't really swarms.' The conclusion lands where it usually does — structure choice gets dragged along by pricing. Part 2 of the series.
The five ways 2026 papers slice an agent, side by side. Two things stand out by the end: Role, Skill, and Judge are different names for the same concept, and the time-axis literature is nearly empty. Part 1 of the series.
What surfaces on the second read of Anthropic's Advisor Tool. This isn't new architecture — it's a temporary fix shaped by 2026 pricing. A pattern that disappears once Opus prices drop, and eleven other papers from the same period are quietly moving the same way. The anchor of the series.
How do you organize the sliced agents. Four structures, paired with one skeptical paper that argues 'LLM swarms aren't really swarms.' The conclusion lands where it usually does — structure choice gets dragged along by pricing. Part 2 of the series.
The five ways 2026 papers slice an agent, side by side. Two things stand out by the end: Role, Skill, and Judge are different names for the same concept, and the time-axis literature is nearly empty. Part 1 of the series.
Reading the 2026 AI Index left a recurring impression — this year's edition lands softer than last year's. This piece chases that impression. Side by side: the report's headline indicators, and the shifts that landed outside it during the eight weeks before publication.
AI products are fundamentally different from traditional software. This post distills 12 counterintuitive lessons from 20 builders, the CC/CD framework for AI development lifecycles, and a 3-phase eval system — all sourced from Lenny Rachitsky's newsletter.
Self-Tuning Loop 4 steps (Generate → Capture → Analyze → Evolve) extracted as a universal module. Supabase DDL, diff capture utilities, analysis/evolution prompts, email/blog examples, GitHub reference implementation.
The actual production system implementing Self-Tuning Loop from Part 1. Data collection (35 sources), AI curation, Telegram input pipeline, weekly auto-review, and Syncthing-based zero-deploy prompt evolution.
Operational review of WICHI's backend on Railway and Supabase, covering real-world incidents like connection pooling and migration conflicts, plus criteria for infrastructure migration.
Analysis of the design philosophy and hierarchical dependencies of the four GEO Score layers (Inclusion, Prominence, Quality, Stability), and their implications for decision-making.
Introducing Self-Tuning Loop: capture implicit feedback from human edit diffs, analyze patterns periodically, and auto-evolve prompt guidelines. Includes academic gap analysis (DSPy, TextGrad, POHF).
Documentation of WICHI's 9-Bucket query framework. Defines 3 Zones and 9 Buckets based on brand presence to measure AI's organic recommendations effectively.
Analysis of multi-engine architecture design principles that leverage response variance as signals, featuring parallel collection structures and scalability via the adapter pattern.
A guide to zero-cost minimum monitoring using Sentry and Betterstack, showing how an 80-minute pre-launch investment reduced error response time from 13 hours to 1 hour.
Documenting the psychological mechanisms and viral loop strategies behind three sharing elements designed for MMU's PLG growth: score cards, badges, and custom checklist links.
Review of SAGEO Arena and CORE papers, analyzing the need for integrated GEO evaluation frameworks and the vulnerability of AI search rankings (91.4% Top-5 manipulation success rate).
Solving the false-pass problem in MMU's checklist by designing a condition marker system with 17 feature flags and improving score accuracy through 48 regression tests.
Three Anthropic announcements across 96 hours in April 2026, broken down into Access, Capability, and Execution layers to reveal a coordinated vertical integration strategy.
Review of AutoGEO (quality-preserving auto-optimization) and E-GEO (e-commerce vertical benchmark) papers, analyzing how GEO optimization seeks a Pareto optimal between visibility and utility.
Review of the KDD 2024 GEO paper and Chen et al. 2025. Covers the academic definition of GEO, the PAWC visibility metric, and AI search's preference for earned media.
OSS monetization framework for solo builders. Outlines a 5-stage strategy using the 'What (Free) - How (Paid)' model to generate revenue without enterprise sales or managed infrastructure.
Strategic analysis of three opportunity factors in the AI search (GEO) market and why WICHI chose 'SaaS-based monitoring' over advertising or agency models.
Analysis of AI infrastructure monetization. Details the 'software free, operations paid' model, where revenue is driven by GPU compute hours and data storage volume in the AI stack.
Analysis of major players entering the GEO market (SearchGPT, Perplexity, Google AI Overviews), their response characteristics, and step-by-step response strategies for enterprises.
6-Layer agent system design canvas and 8+4 core patterns from the Claude Code source leak (512K lines TS). From the Circuit Breaker that stopped 250K/day wasted API calls with 3 lines, to 23-step bash AST security.
The March 31, 2026 npm sourcemap incident revealed Claude Code internals. 4-phase execution, 7 modes, and the 11-step Agent Loop analyzed.
44 feature flags, 20 externally inactive. KAIROS, Auto-Dream, UltraPlan, Coordinator, Bridge, Daemon, UDS Inbox, Buddy, plus anti-distillation and undercover mode.
CC internal Memory System (4-type persistent memory, Auto-Dream, Auto-Compact) vs independently built 3-layer system (documents→index→semantic search). validate_placement() is the differentiator CC doesn't have.
52 built-in tools' common interface, 10-step execution pipeline, safe=parallel/unsafe=sequential concurrency model, 5-stage permission pipeline, and 888KB Tree-sitter 23-step bash security.
Analysis of Langfuse and Dify's open-source monetization. Explains why observability layers are structurally superior to frameworks due to higher switching costs and continuous usage.
Ralph Loop solved context rot but remains prompt-bound. This post maps the trajectory from ALAS autonomous parameter updates to Self-Evolving Agent loops, Multi-Agent Swarms with World Models, Korea's Ralphathon results (100K LOC, 70% tests, zero human keystrokes), and the concrete shift in developer roles from implementation to specification and verification.
Starting from while true + cat task.md, building up through stop hooks, file-based state persistence, and cross-model worker-reviewer separation. Three practical examples — coding migration, prompt refinement, and test coverage expansion — plus analysis of the open-source ecosystem and Korea's Ralphathon.
RLHF, ReAct, Reflexion, LangGraph/AutoGen, Context Rot, Ralph Loop. Six generations of agent loop architecture — what each solved, what each broke, and why a Bash while-loop turned out to be the answer.
Three-scenario analysis (Bull/Base/Bear) of the AI market's $660B CapEx and survival strategies by segment
Based on ChartMogul data, an analysis of AI-native SaaS GRR at 40% -- the wrapper curse and the common traits of companies that broke through it
Analysis of four revenue models in the GEO agency and consulting market, risk mitigation strategies for search engine updates, and the necessity of transitioning to tech-driven SaaS.
A structural analysis of South Korea's AI strengths and weaknesses, the impact of the AI Basic Act, and five areas where the country can realistically compete
Monetization strategies of LangChain, LlamaIndex, and CrewAI. Analysis of the 'free framework, paid operations' pattern and its application to solo builder models.
Dissecting NVIDIA H100's 88% margin structure, analyzing how the AI chip monopoly impacts the ecosystem, and examining 3 scenarios for disruption
Cross-comparing Lex Fridman #490 (2/1) and Noh Jungseok EP91 (3/21) to trace how perceptions of AI agents shifted before and after OpenClaw
First installment of a series analyzing the AI market across three layers: infrastructure ($500B+), platform ($18B+), and applications ($80B+)
Comparative analysis of 10 major GEO SaaS players (Profound, Scrunch, Peec, etc.) covering features, pricing, and targets to derive WICHI's strategic positioning.
Predicting the 5-stage GEO value chain based on 20 years of SEO history, analyzing business opportunities and preemptive strategies for infrastructure, analytics, and specialized media.
Why MMU is an MIT open-source CLI, not a paid SaaS. Details the solo builder workflow fit, the 3-tier 'What-How-Auto' revenue model, and explicit validation metrics.
Documenting how 534 SaaS launch checklist items were derived from 80 service analyses, 12 guidelines, and 5 real-world failures, including the P0-P3 priority logic.
Analyzing the difference between 'code complete' and 'product complete' through the 3-week post-feature work (payment stability, security, legal docs, SEO, etc.) which accounted for 88% of the effort.
10 key changes for transitioning an MVP to commercial SaaS: security hardening, JWT auth, KO/EN i18n, and payment automation to build a 'payment-ready' service.
Recording the 3-day MVP build for WICHI during the Jocoding Hackathon, covering tech stack choices (FastAPI, React, Supabase) and priorities for creating a 'working product' under tight deadlines.
Comparative analysis of ChatGPT Search, Perplexity, and Google AI Overviews on accuracy, speed, and cost, with optimal engine recommendations for each query type.
Definition of GEO and its six structural differences from SEO. Proposes a new framework for brand visibility in an environment where AI search provides answers instead of link clicks.
Analysis of the LLM-as-Judge pattern for evaluating AI response quality, featuring multidimensional metric design, reliability verification, and strategies for position and verbosity bias.
Comparative analysis of Lemon Squeezy (MoR) and Stripe for Korean solo SaaS builders, providing a practical guide on global tax compliance, fees, and business registration.
Recording the 24-hour pivot of WICHI to an independent SaaS after a hackathon rejection, covering i18n implementation, SEO setup, and monetization roadmap restructuring.
Migration from Lovable ($25/mo) to Vercel ($0) including build pipeline reconfiguration, documenting the frontend hosting transition process.
Analyzing n8n's Fair-code license strategy as a defense against AWS risk in open source monetization, and how licensing defines the boundaries of viable business models.
Comparison of open source licenses (MIT, Apache 2.0, BSL, etc.) and a decision framework based on project scale and revenue model, including MMU's choice of MIT.
Essential security checklist for solo SaaS builders. Defense against five critical OWASP Top 10 vulnerabilities and real-world security configuration cases improved for WICHI.
Personal execution notes from a non-tech builder who started with AI FOMO and is now navigating the messy reality of production beyond the initial 'one-click' hype.